DECISION MAKING IN SUPPLY RISK AND SUPPLY DISRUPTION

DECISION MAKING

IN SUPPLY RISK AND

SUPPLY DISRUPTION MANAGEMENT

Inauguraldissertation

zur Erlangung des akademischen Grades

eines Doktors der Wirtschaftswissenschaften

der Universität Mannheim

vorgelegt von

Maximilian Merath

Mannheim

ii

Dekan: Prof. Dr. Dieter Truxius

Referent: Prof. Dr. Christoph Bode

Korreferentin: Prof. Dr. Laura Marie Edinger-Schons

Tag der mündlichen Prüfung: 7. Dezember 2018

iii

Acknowledgements

First of all, I would like to express my sincere gratitude to my supervisor Prof. Dr.

Christoph Bode. Christoph taught me everything I needed to know about conducting

research and presenting one’s work in an adequate fashion. He provided valuable

feedback and guidance throughout the time I spent at his Chair. Most importantly, I highly

appreciate the positive and goal-oriented working atmosphere Christoph established right

from the beginning. This fostered productivity and progress regarding my work on the

dissertation. Moreover, I could always rely on Christoph regarding issues that extended

beyond the usual work-related matters.

Special thanks go to my colleagues, especially Michael Westerburg and Sebastian

Gehrlein, for valuable discussions, fun lunch breaks, and the distractions from the daily

work. In addition, I would like to thank and Helke Naujok and Judith Fuhrmann.

Furthermore, I owe a deep debt of gratitude to my parents, Esther and Franz, my

fiancée, Danica, and my brother, Janosch. I could always rely on their support and

suggestions.

Finally, an exciting part of the time I spent working on the dissertation has been

made possible by John R. Macdonald and Lynn M. Shore who invited me to Colorado

State University (Fort Collins, Colorado). John provided excellent advice on chapter 2 of

this dissertation and played a decisive role in making my stay in Fort Collins a great

experience.

iv

Contents

List of figures ............................................................................................. vii

List of tables .............................................................................................. viii

List of abbreviations .................................................................................. ix

Chapter 1 Introduction and research overview ....................................... 1

1.1 Motivation ......................................................................................................... 1

1.2 Research questions ........................................................................................... 2

1.2.1 Research question 1: Supply risk management ...................................... 3

1.2.2 Research question 2: The influence of supply risk management on the

impact of disruptions .............................................................................. 4

1.2.3 Research question 3: Supply disruption management ............................ 6

Chapter 2 Supply disruptions and protection motivation: Why some

managers act proactively (and others don’t) ......................... 8

2.1 Introduction ...................................................................................................... 9

2.2 Protection motivation theory and propositions ........................................... 10

2.2.1 Coping appraisal ................................................................................... 13

2.2.2 Threat appraisal ..................................................................................... 13

2.2.3 Individual characteristics ...................................................................... 15

2.3 Methodology .................................................................................................... 16

2.3.1 Experimental design ............................................................................. 17

2.3.2 Study participants ................................................................................. 20

2.4 Results .............................................................................................................. 23

2.4.1 Estimation strategy ............................................................................... 23

2.4.2 Model estimation .................................................................................. 25

2.5 Discussion ........................................................................................................ 28

2.5.1 Theoretical implications ....................................................................... 29

2.5.2 Managerial implications ....................................................................... 31

2.5.3 Limitations and future research opportunities ...................................... 32

v

Chapter 3 Substantive and symbolic corporate social responsibility:

Blessing or curse in case of misconduct? .............................. 34

3.1 Introduction .................................................................................................... 35

3.2 Conceptual background and hypotheses development ............................... 36

3.2.1 Substantive and symbolic management of corporate social

responsibility expectations .................................................................... 36

3.2.2 Corporate social irresponsibility ........................................................... 39

3.2.3 Assimilation and contrast effects subsequent to CSIR ......................... 41

3.3 Method ............................................................................................................. 44

3.3.1 Development of vignettes and experimental design ............................. 45


3.3.3 Measures ............................................................................................... 48

3.3.4 Manipulation checks ............................................................................. 50

3.4 Results .............................................................................................................. 51

3.5 Discussion ........................................................................................................ 54

3.5.1 Theoretical implications ....................................................................... 54

3.5.2 Managerial implications ....................................................................... 57


Chapter 4 Supply disruption management: The early bird catches the

worm, but the second mouse gets the cheese? ...................... 61

4.1 Introduction .................................................................................................... 62

4.2 Supply disruption management and resilience ............................................ 63

4.3 How managers should behave in supply disruption response situations .. 65

4.3.1 A model of supply disruption recovery ................................................ 65

4.3.2 The environment ................................................................................... 66

4.3.3 Organizational adaptation to environmental change ............................ 67

4.3.4 Supply disruptions and uncertainty ....................................................... 68

4.3.5 Disruption response strategies .............................................................. 69

4.3.6 Path dependence ................................................................................... 72

4.3.7 Results ................................................................................................... 73

4.3.8 Robustness of simulation results ........................................................... 80

4.3.9 Discussion ............................................................................................. 81

4.4 How managers actually intend to behave in supply disruption response

situations .......................................................................................................... 82

vi

4.4.1 Development of the vignettes ............................................................... 83


4.4.3 Dependent variable ............................................................................... 85

4.4.4 Manipulation checks ............................................................................. 85

4.4.5 Results ................................................................................................... 86

4.4.6 Discussion ............................................................................................. 88

4.5 General discussion .......................................................................................... 89

4.5.1 Implications for research ...................................................................... 89

4.5.2 Implications for practice ....................................................................... 91


Chapter 5 Conclusion and future research directions .......................... 94

5.1 Summary ......................................................................................................... 94

5.1.1 Research question 1: Supply risk management .................................... 94

5.1.2 Research question 2: The influence of supply risk management on the

impact of disruptions ............................................................................ 96

5.1.3 Research question 3: Supply disruption management .......................... 97

5.2 Limitations ...................................................................................................... 98

5.3 Outlook ............................................................................................................ 99

References ................................................................................................ 101

Curriculum vitae ..................................................................................... 120

vii

List of figures

Figure 1: Overview of research questions ....................................................................... 3

Figure 2: Overall model of protection motivation theory (based on Rogers (1983)) .... 12

Figure 3: Relative importance of DCE response alternative variables .......................... 27

Figure 4: Perceived relative importance of DCE response alternative variables ........... 28

Figure 5: Enriched model of protection motivation theory ........................................... 30

Figure 6: Interaction effect of CSR reputation and CSIR severity ................................ 53

Figure 7: Typical supply disruption profile ................................................................... 64

Figure 8: An individual simulation run .......................................................................... 71

Figure 9: Summary of results ......................................................................................... 75

Figure 10: Post-disruption performance (PDP) ............................................................. 76

Figure 11: Final performance (FP) ................................................................................ 78

Figure 12: Uncertainty resolution performance (URP) ................................................. 79

Figure 13: Number of time periods required to reach long-term configuration (NTP) . 80

Figure 14: The effects of response uncertainty, complexity, and path dependence on

ITA ............................................................................................................... 88

viii

List of tables

Table 1: Scenario descriptions ....................................................................................... 19

Table 2: Attribute level descriptions .............................................................................. 20

Table 3: Multi-item measurement scales ....................................................................... 22

Table 4: Bivariate correlations and descriptive statistics ............................................... 22

Table 5: Estimated MNL models ................................................................................... 26

Table 6: Vignette module text descriptions ................................................................... 47

Table 7: Frequencies, cell means, standard distributions, and correlation for the

dependent variables ......................................................................................... 51

Table 8: ANOVA results with purchase intention (PI) as dependent variable .............. 51

Table 9: ANOVA results with intention to engage in negative word-of-mouth (nWOM)

as dependent variable ....................................................................................... 52

Table 10: Summary of hypotheses and results ............................................................... 53

Table 11: Parameter values determining the experimental conditions .......................... 74

Table 12: Frequencies, cell means, and standard distributions for ITA ......................... 87

Table 13: ANOVA results .............................................................................................. 87

Table 14: Summary of propositions ............................................................................... 90

ix

List of abbreviations

ANCOVA Analysis of covariance

ANOVA Analysis of variance

AVE Average variance extracted

BP British Petroleum

B2B Business-to-business

B2C Business-to-consumer

CFA Confirmatory factor analysis

CFI Comparative fit index

CI Confidence interval

CSIR Corporate social irresponsibility

CSR Corporate social responsibility

DCE Discrete choice experiment

df Degrees of freedom

DICO Direct implementation costs

ESG Environmental, social, and governance

FP Final performance

IPT Information-processing theory

ITA Intention to take immediate action

KPI Key performance indicator

M Mean

MNL Multinomial logit

MS Mean square

NL Nested logit

NTP Number of time periods

nWOM Negative word-of-mouth

PDP Post-disruption performance

PI Purchase intention

PMT Protection motivation theory

RAF Ready-aim-fire

RECO Response costs

REFF Response efficacy

RFA Ready-fire-aim

RMSEA Root mean square error of approximation

SD Standard deviation

SE Standard error

SEFF Self-efficacy

SLO Social license to operate

SRMR Standardized root mean square residual

SS Sum of squares

TLI Tucker-Lewis index (also NNFI)

URP Uncertainty resolution performance

VULN Vulnerability

VW Volkswagen

Introduction and research overview 1

Chapter 1 Introduction and research overview

1.1 Motivation

“The ultimate measure of a man is not where he stands in moments of comfort and

convenience, but where he stands at times of challenge and controversy.”

– Martin Luther King, Jr.

Supply chains are exposed to a myriad of risks that pose considerable challenges to

managers. In case a risk materializes and creates a subsequent disruption, the potential

negative consequences are devastating. Recent developments foster the vulnerability of

firms to suffer from supply chain risks and add considerable relevance to the threat of

being severely harmed by them. First, modern supply chains have become more complex

and interconnected due to globalization and intensified competition (Bode & Wagner,

2015; Business Continuity Institute, 2016). Second, many supply chain management

initiatives that appear to provide enhanced efficiency or responsiveness in stable

environments turn out to be a burden for firms in turbulent ones (Norrman & Jansson,

2004). Third, the pressure that stakeholders can exert on firms to comply with ethical

standards and commit to environmental and social values, especially by holding them

accountable for misconduct within their supply chains, has grown (Hartmann & Moeller,

2014). Finally, not only the frequency of natural catastrophes but also their intensity has

increased (Munich Re, 2017).

Prior research has shown that supply chain disruptions tend to be “more critical

when they occur upstream in the chain” (Pereira, Christopher, & Lago Da Silva, 2014, p.

627). Hence, this dissertation research focuses on upstream supply chain risks (hereafter:

Supply risks) and disruptions (hereafter: Supply disruptions). To manage a firm’s

exposure to supply risks and better prepare for supply disruptions, the responsible

managers can employ two basic approaches. On the one hand, they can act proactively

and aim at reducing the probability of a supply disruption to occur or mitigating the

consequences of such adverse events (supply risk management). On the other hand, they

can decide to cope with the aftermath of a materialized risk in a reactive fashion (supply

disruption management) (Craighead, Blackhurst, Rungtusanatham, & Handfield, 2007).

Various proactive and reactive activities have been delineated in prior research and the

negative consequences of supply disruptions are well-known (Hendricks & Singhal,


2005b). However, and although most executives make supply risk management a top

priority, many firms appear to be unprepared to cope with supply disruptions (A.T.

Kearney & RapidRatings, 2018).

Certain types of supply disruptions – especially those characterized by high impact

but low probability – cannot be avoided or resolved as part of daily operations

management. These events need to be appropriately addressed to mitigate their

potentially severe negative consequences, as highlighted by recent industry examples: An

explosion at a steel supplier’s plant in Nagoya, Japan, forced Toyota to halt production at

all of their Japanese factories for one week (Tovey, 2016), online fashion company ASOS

was exposed to massive criticism due to the identification of child workers in its supply

chain (J. Webb, 2016a), and BASF suffered from a natural gas shortage that required a

shutdown of one of its major factories in China (Bradsher, 2017). To effectively prepare

for such sudden disruptions and mitigate their negative repercussions is a complex task.

In this regard, relatively little is known about the behavior of individuals within

supply risk and supply disruption management (Macdonald & Corsi, 2013). Given the

circumstance that a large fraction of the decisions concerning crisis situations is typically

made by single decision makers with centralized authority (Dubrovski, 2004), it is

surprising that behavioral aspects have been neglected in research on supply risk and

disruptions, so far. Thus, the aim of this dissertation research is to shed light on specific

research questions that cover behavioral aspects of managing supply risks and disruptions

to improve our understanding of how to effectively address them. The results not only

provide novel implications for theory and practice but also reveal certain fruitful avenues

for future research.

1.2 Research questions

The research questions addressed in the course of this dissertation revolve around

important issues of decision making in supply risk and supply disruption management.

These issues have received only limited research attention and benefit from novel insights

to improve our understanding of how supply risks and disruptions can effectively be

addressed. In the extant literature on supply risks and disruptions, there is an agreement

that supply disruptions follow a typical profile with regard to their impact on firm

performance over time (Sheffi, 2005). In case that a supply risk materializes, a subsequent


supply disruption leads to a sudden drop in operating performance. This disturbance

causes firms to initiate recovery efforts to return to normal performance levels.

Figure 1 depicts this typical supply disruption profile and provides an overview of

how the three research questions addressed by this dissertation are linked to it. In

particular, the first research question explores why some managers act proactively to

mitigate the potential loss from future supply disruptions while others do not. The second

research question aims to shed light on how prior engagement in corporate social

responsibility (CSR) affects negative stakeholder reactions to a materialized CSR-related

risk. Finally, the third research question addresses the issue of how quickly decision

makers should and do actually initiate recovery efforts after their firm has been hit by a

supply disruption. Each of these research questions was approached by means of carefully

designed and executed experiments to enable a controlled test of the relationships

investigated. In the following, the three research questions are delineated in more detail.

Figure 1: Overview of research questions

Note. CSIR refers to corporate social irresponsibility.

1.2.1 Research question 1: Supply risk management

The overarching aim of supply risk management is to pursue a combination of activities

for which the remaining amount of risk complies with the firm’s risk preference and

corporate strategy (Hofmann, Busse, Bode, & Henke, 2014). Although it is well-known

that supply risk management is associated with certain benefits (Mitroff & Alpaslan,

2003; Norrman & Jansson, 2004), it remains largely unexplored how and why some


managers decide to take proactive action to mitigate the impact of future supply

disruptions while others do not. The related managerial choices strongly influence to

which degree firms are able to cope with supply disruptions. Hence, to contribute to an

improved understanding of this issue, it is vital to unravel behavioral components of

supply risk management.

A relatively similar problem has been investigated by health-related research.

Analyzing the factors that shape an individual’s decision to adopt certain preventive

health behaviors, Rogers (1975, 1983) developed protection motivation theory (PMT).

PMT is based on the idea that when individuals are exposed to a threat, their probability

to adopt a certain coping response depends on two cognitive appraisal processes. These

appraisal processes take factors into account that are associated with the costs and impact

of a potential proactive action (coping appraisal) as well as the characteristics of a threat

and the consequences of not taking proactive action (threat appraisal). Although PMT

has initially been developed to study the effects of fear appeals on health behavior, it has

moved far beyond them and is considered generalizable to “apply to any situation

involving threat” (Rogers, 1983, p. 172).

PMT serves as an insightful framework to study decisions on whether or not

managers proactively address supply risks and contributes to a better understanding of

the role that individual managers’ behaviors play within the process of managing supply

risks. Hence, Study 1 in Chapter 2 employs PMT to examine proactive decision making

to provide an answer to the following research question:

Research

Question 1

Why do some decision makers decide to proactively act to mitigate

the impact of future supply disruptions while others do not?

1.2.2 Research question 2: The influence of supply risk management on the impact

of disruptions

Despite the growing pressure that stakeholders can exert on firms to adhere to ethical

standards and behave in socially responsible ways (Campbell, 2007), the amplified public

awareness and availability of information on environmental and social misconduct

(Fiaschi, Giuliani, & Nieri, 2017), and the increasingly difficult challenge for firms to

prevent irresponsible behavior in their supply chains (Hartmann & Moeller, 2014), the

issue of corporate social irresponsibility (CSIR) in supply chains has been neglected in

the academic discussion of CSR. In the academic literature on CSR, there is a strong focus

on equating CSR with “doing good”, although it has been shown that “avoiding bad” also


constitutes an important precondition for a firm to successfully position itself as socially

responsible (Lin-Hi & Müller, 2013).

Prior research has predominantly assumed that a firm’s ex ante CSR activities may

serve as a “reservoir of goodwill” among the firm’s stakeholders that mitigates negative

reactions to CSIR (e.g., Flammer, 2013; Godfrey, Merrill, & Hansen, 2009; Jones, Jones,

& Little, 2000; Klein & Dawar, 2004). However, there are reasons to seriously doubt that

these insurance-like effects of ex ante CSR in case of CSIR universally apply. Several

studies suggest that firms which engage in CSR experience more negative reactions to

CSIR compared to firms which do not promote themselves as socially responsible (e.g.,

Sen & Bhattacharya, 2001; Swaen & Vanhamme, 2003; Vanhamme & Grobben, 2009).

Moreover, examples like Volkswagen (VW) that won numerous awards for CSR but

recently faced severe criticism due to the “Dieselgate” scandal contradict the idea of

insurance-like effects of prior CSR (Lynn, 2015).

Some researchers argue that a possible explanation for these contrary effects might

be that, similar to brand commitment (Germann, Grewal, Ross, & Srivastava, 2014), a

reputation for CSR provides a goodwill buffer in case of non-severe CSIR but also serves

as an expectation burden in case of severe CSIR (e.g., Janssen, Sen, & Bhattacharya,

2015; Kang, Germann, & Grewal, 2016). Moreover, these effects of ex ante CSR

reputations in times of crises might be subject to the nature of the employed activities to

gain a CSR reputation, which can either be substantive or symbolic (Ashforth & Gibbs,

1990). Substantive CSR is typically perceived as intrinsically motivated while symbolic

CSR is associated with extrinsic motives. In case of CSIR, instead of assuming a lack of

proper management or even malevolence, stakeholders are more willing to develop

alternative explanations for CSIR if they perceive a firm’s CSR engagement to be

intrinsically motivated rather than extrinsically driven (Janssen et al., 2015).

Study 2 focusses on the role of the purchasing function as gatekeeper to CSIR-

related risks in the supply chain of a focal firm and contributes to a better understanding

of negative stakeholder reactions to CSIR. More specifically, this research investigates

whether the effect of a CSR reputation on negative stakeholder reactions to CSIR in

professional buying contexts depends on CSIR severity and the reputation’s nature

(substantive vs. symbolic). Thereby, we explore the influence of proactive CSR

engagement on negative stakeholder reactions to a realized CSIR-related risk. Thus, the


research presented in Chapter 3 aims to provide an answer to the following research

question:

Research

Question 2

How are negative stakeholder reactions to CSIR in industrial

buying contexts shaped by the type of CSR reputation and the CSIR

severity?

1.2.3 Research question 3: Supply disruption management

Decision making in supply disruption response and recovery situations is often difficult,

because choices have to be made dynamically in complex environments that are

characterized by uncertainty and limited information. Under these conditions, there are

two basic approaches of taking recovery decisions. Some firms defer actions until reliable

information is available for a sound judgment; other firms respond immediately, even

though the information at hand is cloudy or fragmented (Kleinmuntz & Thomas, 1987).

Using the analogy of shooting, the first approach can be termed “ready-aim-fire” (RAF)

and the second “ready-fire-aim” (RFA). Intuitively, RAF may lead to more precise

actions and a better solution quality than RFA, but the time required to gather additional

information may allow quicker competitors to obtain superior positions. Conversely,

RFA carries the inefficiencies of trial and error and, when decisions are irreversible and

path-dependent, the risk of pursuing an inferior recovery path. Hence, the key question

for managers concerned with supply disruptions is: Under which conditions should a

decision maker delay its recovery decision until more evidence is acquired?

It is not always trivial to select one of the two presented approaches (RAF and RFA)

as highlighted by the often-cited Albuquerque fire (Latour, 2001). On March 17, 2000, a

Philips plant in Albuquerque, New Mexico, was hit by lightning and caught fire (Lee,

2004). Initially, it seemed that the damage would be limited. Philips informed the two

main customers of the chips produced at this plant, Nokia and Ericsson, about an expected

delivery delay of one week. Yet, the responses of the two competitors were completely

different. Nokia immediately put pressure on Philips and quickly collaborated with

alternative suppliers to recover from the disruption with merely limited negative

consequences. Ericsson adopted a “wait and see” approach and delayed remedial action

until more evidence about the supply disruption had been acquired. By the time it became

obvious that the damage to the clean rooms was far more severe than expected, Ericsson

was not able to find an alternative supplier on short notice. As a result, Ericsson had to


delay the launch of a new product and, finally, also ceased its own handset production as

a further consequence of this incident (Latour, 2001; Sheffi, 2005).

This example highlights that a firm’s ability to effectively respond to supply

disruptions is not only vital to its short-term performance but also essential for its long-

term competitiveness. However, supply disruption management has received only limited

research attention. In particular, there is a need to gain a better understanding of how the

responsible managers should respond to disruptive events in their firm’s upstream supply

chain to quickly recover from their negative consequences (Bode, Wagner, Petersen, &

Ellram, 2011; Macdonald & Corsi, 2013). Moreover, there is a need to investigate how

and to which extent the responsible decision makers’ intended choices deviate from how

they should behave. Hence, the following research question is addressed by Study 3 as

presented in Chapter 4:

Research

Question 3

Under which conditions should a manager delay its response

actions until more evidence is acquired and how do managers

actually intend to behave?

Supply disruptions and protection motivation: Why some managers act proactively (and others don’t) 8

Chapter 2 Supply disruptions and protection

motivation: Why some managers act

proactively (and others don’t)

Co-authors:

Christoph Bode

Endowed Chair of Procurement, Business School, University of Mannheim, Germany

John R. Macdonald

Department of Management, Colorado State University, Fort Collins, USA

Abstract1

Supply disruptions present considerable managerial challenges and have severe

consequences. To protect their firms from disturbances, managers must decide whether

or not to take proactive measures. However, little is known about how they make these

decisions. Protection motivation theory suggests that an individual’s intention to respond

to a threat by acting proactively results from cognitive appraisal processes. These

processes evaluate the characteristics of a potential coping response (e.g., its effectiveness

in averting the threat) and the threat itself (e.g., its severity). Building on this framework,

this study presents an analysis of what drives managers to or deters them from responding

to the threat of a supply chain disruption. The exploratory results from a discrete choice

experiment show that decision makers have a strong subconscious focus on cost-related

aspects of a specific proactive action, while consciously prioritizing the efficacy of the

action over its costs. Thus, the study provides interesting and novel insights into

behavioral aspects of supply chain risk management by revealing that decision makers’

perceptions of the relative importance of proactive action attributes deviate considerably

from their actual choice behavior. Additionally, proactive personality, risk attitude,

control appraisal, and experience had significant effects on the relative importance of

certain proactive action attributes.

1 Merath, M., Bode, C., and Macdonald, J. R., 2018. Supply disruptions and protection motivation: Why

some managers act proactively (and others don’t). Unpublished Working Paper, 1-38. An earlier version

was nominated for the “ISM Best Paper in Supply Chain Management Award” of the Operations and

Supply Chain Management Division of the Academy of Management at the Annual Meeting in Chicago,

IL, in 2018.


2.1 Introduction

Supply chain disruptions are “unplanned and unanticipated events that disrupt the normal

flow of goods and materials within a supply chain […] and, as a consequence, expose

firms within the supply chain to operational and financial risks” (Craighead et al., 2007,

p. 132). Minor disruptions can often be resolved within day-to-day operations, but high

impact–low probability disruptions pose serious challenges for managers. Moreover,

disruptions tend to be “more critical when they occur upstream in the chain” (Pereira et

al., 2014, p. 627). For example, BMW suspended production in China for one week due

to supplier parts shortages (Moriyasu, 2017), BASF’s U.S. plant was temporarily shut

down following a disruption at an external supplier (Burger & Sheahan, 2018), and

Toyota suffered from the interruptions in the supply of raw materials after a major

earthquake in Japan (Tajitsu & Yamazaki, 2016). Hence, this study examines disruptions

in a focal firm’s upstream supply chain that cannot be resolved within daily operations

management (hereafter: Supply disruptions).

There are two ways to address and manage a firm’s exposure to such disruptions.

Managers can either decide to proactively tackle supply disruptions with measures aimed

at minimizing the probability of their occurrence or mitigating their damage (hereafter:

Supply risk management), or reactively cope with the adverse effects of a materialized

risk (hereafter: Supply disruption management) (Craighead et al., 2007). The overall goal

of both approaches is to determine a mix of activities for which the remaining amount of

risk is in line with the firm’s risk preference and corporate strategy (Hofmann et al.,

2014). The related choices considerably affect the degree to which firms can recover from

supply disruptions (e.g., Habermann, Blackhurst, & Metcalf, 2015; Yildiz, Yoon, Talluri,

& Ho, 2016), but behavioral aspects of supply risk management remain relatively unclear

(Bode & Macdonald, 2017). More specifically, although the negative consequences of

supply disruptions (Hendricks & Singhal, 2005a,b) and the benefits of supply risk

management are well-known (Mitroff & Alpaslan, 2003; Norrman & Jansson, 2004),

little is known about why some decision makers proactively act to mitigate the impact of

future supply disruptions while others do not.

Health-related research faces a similar problem. Analyzing the factors that shape

the decision to engage in a certain preventive health behavior, Rogers (1975, 1983)

developed protection motivation theory (PMT). PMT centers upon two cognitive

appraisal processes that account for the impact and costs of a potential proactive action


(coping appraisal) as well as the components of a threat and the consequences of not

taking proactive action (threat appraisal). These appraisal processes ultimately affect a

person’s propensity of adopting a coping response. Although PMT was developed to

discern the effects of fear appeals (appeals using fear as the driving motivation) on health

attitudes and behavior, it has moved far beyond them and is considered applicable “to any

situation involving threat” (Rogers, 1983, p. 172).

This study applies PMT to understand why managers do or do not decide to take

proactive action in preparation for supply disruptions. More specifically, an extended

version of the PMT framework is proposed which also integrates and accounts for certain

individual characteristics that have been identified to influence proactive behavior. The

proposed model was subjected to empirical scrutiny by means of discrete choice

experiments, which emulated a decision making situation under uncertainty and the threat

of a future supply disruption. In order to explore which factors predict managers’ intent

to take proactive action, the participating managers were provided with systematically

manipulated choice scenarios and response options. The empirical results reveal the

relative importance of specific proactive action attributes, highlight a mismatch between

managers’ choice behavior and perceptions, and show relationships between individual-

specific characteristics and proactive action attributes. Thereby, this study contributes to

a better understanding of managers’ behavior when managing supply chain risks.

2.2 Protection motivation theory and propositions

There is consensus among scholars that proactive behavior involves “anticipatory action

that employees take to impact themselves and/or their environments” (Grant & Ashford,

2008, p. 8). Two characteristics that distinguish proactive from reactive behavior are

acting in advance and intended impact (Grant & Ashford, 2008). Proactivity means being

future-focused by planning and selecting specific measures to modify the environment

before certain events occur (Bandura, 2006). Moreover, proactive behavior is goal-driven

and intended to bring about environmental change (Bateman & Crant, 1993).

To better understand an individual’s proactive motivation, most attention has been

devoted to how humans assess the likely outcomes of their behavior (S. K. Parker,

Williams, & Turner, 2006) and the reasons for them to strive for a certain proactive goal

(Griffin, Neal, & Parker, 2007). In the supply risk context, the purpose of taking proactive

action is to protect the focal firm from future damage. Numerous studies, mostly in the


field of health, have been conducted to understand how people decide to behave when

exposed to various threats (e.g., Grothmann & Reusswig, 2006; Sheeran, Harris, & Epton,

2013). In line with these research efforts, several cognitive behavioral theories attempt to

explain how proactive behavior is initiated, but they vary with regard to the assumed

mediating processes and their applicability to different contexts. All major cognitive

behavioral models originated in expectancy-value theories (Hovland, Janis, & Kelley,

1953). The premise of the family of expectancy-value theories is that an individual’s

intention to adopt a specific behavior depends on his or her expectations of its

consequences and its value. Expectancy-value theories were used to shed light on the

effects of fear appeals, which are “persuasive messages with the intent to motivate

individuals to comply with a recommended course of action through the arousal of fear

associated with a threat” (A. C. Johnston & Warkentin, 2010, pp. 550-551).

Three stimuli form a typical fear appeal. The first constitutes a value component

and the other two shape expectations: (1) The magnitude of an event’s harmfulness (value

component), (2) the probability that this event will occur given that no protective behavior

is adopted or existing behavior is modified (expectancy component), and (3) the

availability and effectiveness of a coping response that might prevent adverse

consequences of the event (expectancy component).

Based on expectancy-value theories, PMT was originally developed to study the

impact of fear appeals on health-related behavior and support their effective design

(Rogers, 1975). Its revised version (Rogers, 1983) has enabled the theory to evolve and

encompass not only health-related threats – where it has been successfully applied (Floyd,

Prentice‐Dunn, & Rogers, 2000) – but also nuclear actions (Axelrod & Newton, 1991),

and, more recently, technological or environmental hazards (Boss, Galletta, Lowry,

Moody, & Polak, 2015; Y. Chen & Zahedi, 2016). PMT has become “sufficiently broad

to apply to any situation involving threat” (Rogers, 1983, p. 172) and is considered “an

established, robust theoretical foundation for the analysis and exploration of

recommended actions or behaviors to avert the consequences of threats” (A. C. Johnston

& Warkentin, 2010, p. 552). In line with this, we suggest that PMT offers an adequate

and insightful framework to improve our understanding of how the threat of an impending

supply disruption translates into proactive action. More specifically, it allows us to

investigate the drivers of, and impediments to, an individual’s motivation to take

proactive action. PMT has received widespread empirical support (A. C. Johnston &


Warkentin, 2010) and provides an effective theory for human behavior which overcomes

many conceptual problems leading to low correlations between attitudes and behavior

(McGuire, 1985).

Figure 2 depicts the revised version of PMT (Rogers, 1983), which assumes that

several sources of information about an impending threat, categorized as either

environmental or intrapersonal, may initiate PMT’s key mediating processes. These

sources of information comprise, for example, prior experience or observational learning.

Both prior experience and observational learning are considered important triggers of

adaptation processes in supply disruption management (Bode et al., 2011; Hora &

Klassen, 2013). In addition, it is important to note that “any source of information can

lead to any of the mediating processes” (Rogers, 1983, p. 167).

Figure 2: Overall model of protection motivation theory (based on Rogers (1983))

The mediating processes at the core of PMT are coping appraisal and threat

appraisal. The former evaluates the adaptive response (e.g., taking proactive action to

prevent damage), while the latter assesses the maladaptive response (e.g., maintaining the

status quo). Each process comprises variables that either increase or decrease the

probability of adopting the respective response. As shown in Figure 2, the evaluation of

the variables within each appraisal process is assumed to summate algebraically into a

final appraisal of coping as well as threat; a protection motivation. Finally, both processes

affect the strength of this motivation, which determines whether or not proactive action

is taken. Protection motivation can be understood as a behavioral intention that may result

in a single act, repeated acts, multiple acts, or repeated multiple acts that involve either

direct action or its inhibition. Feedback from coping behavior will enter the PMT model

as “prior experience” for reappraisals of threats and coping responses to adjust an

individual’s protection motivation (Rogers & Prentice-Dunn, 1997).

Sources of

information Cognitive mediating processes Coping modes

Environmental

- Verbal

persuasion

- Observational

learning

Intrapersonal

- Personality

variables

- Prior experience

Action or inhibition

of action

- Single act

- Repeated acts

- Multiple acts

- Repeated,

multiple acts

Factors affecting response probability

Maladaptive

response

Adaptive

response

Increasing Decreasing

Intrinsic rewards

Extrinsic rewards

Severity

Vulnerability

Response efficacy

Self-efficacyResponse costs

=

=

Protection

motivation

Fear

arousal

Threat

appraisal

Coping

appraisal


2.2.1 Coping appraisal

An adaptive response aims at proactively coping with the potential negative consequences

associated with a threat. The coping appraisal process evaluates the ability of an

individual to cope with and avert being harmed by the threat. Three major components

lead to the overall evaluation of coping: Response efficacy and self-efficacy increase an

individual’s willingness to perform a proactive action; the costs associated with the

response decrease it. Response efficacy is the perception that a specific proactive action

will avert the dangers of a threat. Self-efficacy is the belief in one’s ability to successfully

perform a specific action. It was proposed to be a major component of almost all processes

of psychological change and exerts a considerable influence on whether or not a certain

behavior is chosen and how much effort will be invested in its execution (Bandura, 1977).

However, the concept of self-efficacy had largely been neglected in any of the social-

cognitive behavioral models based on expectancy-value theories. The revised version of

PMT was the first of these models to incorporate self-efficacy into the study of proactive

decision making behavior (Rogers, 1983). Finally, any costs of adopting the adaptive

response refer to its response costs which will reduce the appeal of a specific response

option. Response costs can refer to financial costs as well as inconvenience, time, and

effort (Grothmann & Reusswig, 2006).

Hence, the coping appraisal process of PMT suggests that decision makers assess

the appeal of a specific proactive action alternative on the basis of three characteristics:

(1) The effectiveness of the alternative (response efficacy), (2) an individual’s ability to

successfully perform the action (self-efficacy), and (3) the costs associated with taking

the proactive action alternative (response costs). Formally:

Proposition 1. Individual decision makers select proactive action alternatives to

prepare for the threat of an impending supply disruption based on

(a) an alternative’s effectiveness,

(b) perceived self-efficacy, and

(c) the costs associated with an alternative.

If an alternative offers an attractive package in terms of these

criteria, it is more likely to be chosen.

2.2.2 Threat appraisal

A maladaptive response exposes an individual or a firm to a threat (e.g., being

“unprepared” for a supply disruption), and is composed of intrinsic rewards, extrinsic

rewards, severity, and vulnerability (see Figure 2). Intrinsic rewards and extrinsic


rewards increase the probability of choosing a maladaptive response. In contrast, the

severity of a threat and the expectancy of being exposed to the threat (vulnerability)

reduce the attractiveness of a maladaptive response. Severity is the potential amount of

physical or economic damage associated with a threat (Rogers & Prentice-Dunn, 1997).

At the same time, although fear was initially seen as an essential mediating variable of

the effect of fear appeals on behavior, it is not considered to have a direct influence on

protection motivation. It merely has an indirect impact through the evaluation of a threat’s

severity; therefore, fear is treated as “an insignificant byproduct of threat appraisal”

(Tanner, Hunt, & Eppright, 1991, p. 37) to underline the “importance of cognitive

processes rather than visceral ones” (Rogers, 1983, p. 169). PMT posits that the

motivation of an individual to take a maladaptive response will decrease if the threat is

severe, the vulnerability to the threat is high, one is able to successfully perform a

proactive action, and this response can effectively avert the threat’s potential negative

consequences. A maladaptive response’s likelihood will increase if the response is

accompanied by rewards and performing the proactive action is costly.

Prior research using the PMT framework revealed serious difficulties in

operationalizing the rewards of a maladaptive response as distinct from the costs of a

proactive action. For this reason, the vast majority of PMT research has neglected the

rewards component and response costs have to date not been extensively researched.

Although these components are delineated as conceptually different (Rogers, 1983), the

distinction between them might not be clear to managers who could perceive them as

equal (maladaptive response rewards could be perceived as avoiding the costs of a

proactive action). In line with prior research, to avoid duplicating the same factor, we

focus on response costs (Grothmann & Reusswig, 2006; A. C. Johnston & Warkentin,

2010; Tanner et al., 1991).

The vulnerability to a specific threat may affect the relative importance of response

costs. In other words, the vulnerability of a firm to a specific supply disruption might

moderate the influence of a proactive measure’s costs on the decision whether to take

action. If a proactive measure is very costly, a high vulnerability to an impending

disruption may also increase intentions to act proactively due to the higher expected loss

associated with the materialized risk (Knemeyer, Zinn, & Eroglu, 2009; Norrman &

Jansson, 2004). Thus, we propose the following:


Proposition 2. The relative importance of response costs for the probability of

choosing a specific proactive action to cope with a supply disruption

is reduced by the vulnerability of a firm to a supply disruption.

Although PMT presents the two appraisal processes in an unordered fashion, later

studies suggest that threat appraisal precedes coping appraisal (Scherer, 1984, 1988;

Tanner et al., 1991). We follow the notion of an ordered sequence between the two

mediating processes in our experimental design, which will be explained in the

methodology section.

2.2.3 Individual characteristics

Finally, certain characteristics of individual decision makers and of the situation arguably

influence proactive behavior in organizations (S. K. Parker et al., 2006). Based on the

results of a cross-cultural longitudinal study that revealed several key antecedents of

proactive behavior, Frese and Fay (2001) theorized that those antecedents associated with

individual decision makers can be categorized as (1) personality, (2) orientations, and (3)

knowledge, skills, and abilities.

First, personality refers to individual differences that represent proclivity for action

and cross-situational tendencies that activate decision makers. It has been argued that

proactive personality is the most relevant individual predictor of proactive behavior. It

is defined as a disposition toward taking action to bring about change and influence the

environment (Bateman & Crant, 1993; S. K. Parker et al., 2006). In addition, since this

study investigates a decision involving risk, we argue that an individual’s risk attitude

specifies how important a certain risk is to an individual and has a strong influence on

his or her proactive decision making behavior (Heckmann, Comes, & Nickel, 2015).

Different perceptions of the importance of a certain risk may considerably influence

decision making processes and resulting outcomes. Generally, three manifestations of a

decision maker’s risk attitude are distinguished: Risk-averse, risk-neutral, and risk-

seeking (Weber & Milliman, 1997). Prior research has shown that, depending on their

risk attitude, different decision makers may perceive the same risk situation quite

differently which has repercussions on choice behavior (March & Shapira, 1987).

Second, in contrast to the cross-situational and rather general personality factors,

orientations are “behavior tendencies of moderate situational specificity” (Fay & Frese,

2001, p. 106). Orientations motivate proactive behavior by making a person believe that


such behavior is possible and can produce the desired results. Frese and Fay (2001)

proposed that an important orientation for promoting proactivity is a person’s control

appraisal: An individual’s expectation of his or her impact on work outcomes. Decision

makers who assume that their own decisions have a strong effect on their work outcomes

are more likely to engage in proactive action while low levels of control appraisal inhibit

proactive action (Aspinwall & Taylor, 1997).

Finally, knowledge, skills, and abilities capture an individual’s capacity to identify

work-related challenges, analyze them, and develop appropriate solutions (Hunter,

1986). Hence, a person’s understanding of a task determines the ability to act proactively.

As an indicator of an individual’s knowledge, skills, and abilities, experience

considerably affects intention to act proactively. Work experience has been identified as

a main driver of knowledge, and hence, of the development of skills and techniques that

improve job performance (F. L. Schmidt, Hunter, & Outerbridge, 1986). Following prior

research, more experienced decision makers might be more likely to act proactively,

because they possess the requisite knowledge and skills to successfully engage in

proactive behavior (Grant & Ashford, 2008; Grant, Parker, & Collins, 2009).

Based on the framework developed by Frese and Fay (2001), it is proposed that

these four characteristics affect proactive behavior by influencing the relative importance

of PMT’s coping appraisal components:

Proposition 3. The relative importance of PMT’s coping appraisal components for

the probability of choosing a specific proactive action is influenced

by characteristics of individual decision makers, namely their

(a) proactive personality,

(b) risk attitude,

(c) control appraisal, and

(d) experience.

2.3 Methodology

To evaluate the developed propositions and analyze the factors influencing decisions to

engage in proactive preparation for supply disruptions, a discrete choice experiment

(DCE) was developed. DCEs are considered an effective way to analyze complex

decision making tasks and choice behavior (Louviere, Hensher, & Swait, 2000; Moore,

Gray-Lee, & Louviere, 1998). Although they have successfully been applied to analyze

choice behavior in fields such as marketing, economics, or health research, they have only

rarely been used in operations management (e.g., E. J. Anderson, Coltman, Devinney, &


Keating, 2011; Coltman & Devinney, 2013; Pullman, Verma, & Goodale, 2001; Verma,

Louviere, & Burke, 2006). DCEs expose participants to multiple choice situations with

at least two possible response alternatives. Each specific choice situation (comprised of

several choice alternatives) is referred to as choice set. The response alternatives in a

choice set consist of a set of attributes. If no alternative option outperforms the others on

all attributes, decision makers must perform trade-offs between these observed

characteristics of response alternatives. This is described in further detail in the

experimental design section. Regardless of whether these trade-offs are determined

consciously or subconsciously, stated choice preferences reveal the underlying weight or

importance assigned to specific attributes (E. J. Anderson et al., 2011).

2.3.1 Experimental design

The developed experimental design exposed all participants to nine choice sets

comprising three possible response options. The order of choice sets presented was

randomized for each participant to control for order effects (Potoglou & Kanaroglou,

2007). For each choice set, two of the presented alternatives were generic proactive

response actions: “Proactive action A” and “proactive action B”. These alternatives varied

along one or more attributes of interest. Since decisions involving supply risk in practice

also allow individuals to refrain from taking proactive action, a third “no choice”

alternative labelled “neither” was included.2 Whether to include an opt-out alternative is

an important methodological issue and is becoming the norm in choice experiments (J.

R. Parker & Schrift, 2011). Failure to include a “no choice” option may distort the results

by overestimating participation by forcing some participants to choose (Boyle, Holmes,

Teisl, & Roe, 2001). In addition, offering an opt-out option improves the realism of

experiments (Louviere et al., 2000).

A D-optimal3 design allows for an analysis of the attributes’ main effects and the

proposed interactions. To ensure participants’ understanding of the discrete choice

format, an additional tenth choice set was included as a consistency check (Green &

2 We also included the following description to add a more nuanced and realistic notion of what choosing

the “no choice” option actually meant: “Neither, because none of the other response alternatives seem

appropriate.” 3 D-optimality (or D-efficiency) is a widely used and well-established metric in the design of choice

experiments. In order to construct statistically efficient designs, D-optimal designs maximize the Fisher

information matrix (the determinant of the variance-covariance matrix of the model to be estimated)

(Hensher, Rose, & Greene, 2005).


Gerard, 2009). In this tenth set, one of the two proactive actions was clearly constructed

as the dominant alternative (i.e., all attribute levels were more desirable). Thus, to “pass”

the consistency check, respondents had to choose either the dominant or the “no choice”

alternative. At the end of the experiment, participants responded to a brief survey to

collect individual-specific data.

Scenario Design.4 To ensure that the participants were provided with the same

contextual information, they were given a carefully designed introductory paragraph to

read before they were shown the choice sets. This common module delineated the

underlying choice scenario from the perspective of a third person to limit demand

characteristics and effects of social desirability (Fisher, 1993; Thomas, Thomas, Manrodt,

& Rutner, 2013). In this scenario, a procurement manager observes that an earthquake in

Asia has caused a serious supply disruption to a competitor. This manager also sources

parts from Asia so his or her firm is vulnerable to a similar supply disruption. Hence, the

manager needs to decide whether or not to mitigate future potential losses by taking

proactive action. By framing the choice situation in a vicarious learning context, we

account for the relevance of observing other firms as an important source of information

in supply risk and disruption management and increase the experiment’s realism (Hora &

Klassen, 2013). Moreover, supply disruptions due to natural disasters frequently inflict

substantial damage on firms involved (e.g., Helft & Bunkley, 2011; Tajitsu & Yamazaki,

2016; J. Webb, 2016b).

The description of the underlying choice situation discusses the threat appraisal

variables of PMT. First, the potential damage of disruption caused by an earthquake was

described as severe (severity). This is important, because of the focus on high-impact

supply disruptions that cannot be resolved in daily operations management. Second, we

systematically manipulated the probability that the disruption will harm the decision

maker’s firm (vulnerability). We distinguished low from high vulnerability by varying

the earthquake-proneness of the Asian supplier’s location in the description. Participants

were randomly assigned to one of these two treatment conditions to assess whether or not

vulnerability moderates the effect of response costs on choice behavior. In this way we

4 Although DCEs and vignette-based experiments pursue different aims (DCEs: Interested in trade-offs

between attributes; Vignette-based experiments: Investigate the impact of certain variables on observed

intentions or actual behavior), both approaches share similarities with regard to how information about

an a priori defined role and/or situation is presented to respondents. Hence, scenario design and validation

was partly based on recommendations for vignette-experiments (e.g., Aguinis & Bradley, 2014;

Rungtusanatham, Wallin, & Eckerd, 2011).


eliminate systematic differences in the respondents to be able to attribute differences in

response behavior to the treatment condition (Bachrach & Bendoly, 2011). The scenario

descriptions can be found in Table 1.

To assess the realism of the developed scenario and whether the participants

perceived both levels of vulnerability and the potentially severe impact of the impending

disruption as intended (Wason, Polonsky, & Hyman, 2002), we pre-tested the DCE with

72 graduate students. Responses to manipulation checks resulted in significantly different

mean responses for low and high levels of vulnerability. Moreover, high severity was

appropriately represented and the two scenario descriptions with varying degrees of

vulnerability appeared realistic with means of 5.29 and 5.56 (grand mean is 5.44),

evaluated on a 7-point Likert-type scale (1 := “not at all” to 7 := “completely”). The results

of the DCE task were analyzed and used as priors for the generation of the D-optimal

design of the DCE.

Table 1: Scenario descriptions

Introduction

and severity

Leo is procurement manager for Eletrox and is responsible for buying microchips.

Two weeks ago, Eletrox’s major competitor was severely hit by a supply disruption

in Asia that caused this firm to cease its own production for three days and suffer

immense losses. An earthquake destroyed large parts of its supplier’s production

facilities and inventories. In order to prepare for comparable future events, this

competitor subsequently implemented proactive measures.

Factor Manipulated factor levels

Vulnerability

Low High

Eletrox also buys microchips from Asia

and could be severely hit by such a

disruption. Leo considers taking

proactive action to mitigate future losses.

Eletrox’s microchip supplier is located in

an only slightly earthquake-prone

region. Hence, its vulnerability to

earthquake-related supply disruptions in

Asia is low.

Eletrox also buys microchips from Asia

and could be severely hit by such a

disruption. Leo considers taking

proactive action to mitigate future losses.

Eletrox’s microchip supplier is located in

a very earthquake-prone region.

Hence, its vulnerability to earthquake-

related supply disruptions in Asia is high.

Attributes and Their Levels. In the early design stages of a DCE, relevant attributes

for the decision task need to be identified. The coping appraisal process of PMT suggests

that decision makers assess the appeal of a specific proactive action alternative based on

three characteristics: (1) The effectiveness of the response alternative in averting or

mitigating a threat (response efficacy), (2) an individual’s ability to successfully perform

the action (self-efficacy), and (3) the costs of implementing the potential proactive action

alternative (response costs). To appropriately represent these characteristics without

making the task too complicated for the participants, each variable was operationalized


as binary with a low and a high level, by means of appropriate descriptions. Since prior

research indicates that proactive measures are not only accompanied by direct financial

costs but also have the potential to harm the relationship with a supplier (e.g., Heide,

Wathne, & Rokkan, 2007; Zsidisin & Ritchie, 2008), we distinguished two types of

response costs: Direct implementation costs and negative side effects on the respective

buyer-supplier relationship (relationship costs). For instance, if a supplier formerly used

as a single source for a specific part loses a considerable fraction of the demand to a

second source because the buying firm seeks to reduce its risk associated with single

sourcing, the original supplier might be less willing to develop new innovations for this

customer “because of a smaller possibility to amortise the expenses” (Zsidisin & Ritchie,

2008, p. 131). Accordingly, we describe relationship costs as decreased investments of

the supplier into innovations for the buying firm. Table 2 offers descriptions of attributes’

levels.

Table 2: Attribute level descriptions

Attributes Attribute levels

Low High

Response efficacy This action can, to a small extent,

reduce potential future losses

This action can, to a large extent,

reduce potential future losses

Self-efficacy Leo doubts that he will be able to

successfully implement this action

Leo is sure that he will be able to

successfully implement this action

Direct implementation

costs

Direct implementation costs for this

action are comparatively low

Direct implementation costs for this

action are comparatively high

Relationship costs

The current microchip supplier will

spend slightly less money to develop

innovations for Eletrox

The current microchip supplier will

spend considerably less money to

develop innovations for Eletrox

2.3.2 Study participants

Data were collected between January and March 2018 by means of a self-administered

online survey. In total, 308 professionals with direct experience in supply chain

management from Germany, Austria, and Switzerland, were invited to participate.

Contact addresses were obtained from a commercial business data provider. The

managers received an invitation via email and were randomly assigned to one of the two

vulnerability conditions. Of the 308 professionals, 133 completed the DCE, resulting in

an effective response rate of 43.2%. On average, participants had almost 15 years of

experience in supply chain management (SD = 12.52). The participants were able to

choose between a German and English version of the survey. Thirty-three of the 133

participants (26.3%) opted for the English language version. To validate translation


equivalence, the complete experimental material was carefully translated into German by

native speakers and then back-translated into English to ensure equivalent meaning (Craig

& Douglas, 2005). Each participant responded to nine choice sets (and an additional tenth

choice set as a consistency check). The responses of one participant were excluded due

to unrealistically short participation duration. Further data were excluded because nine

participants consistently chose the same alternative, resulting in a full sample of 1107

observations. Hence, the data set comprises 123 participants (58 of these completed the

low-vulnerability condition).

Individual Characteristics. After the choice sets, the participants were asked to

respond to several survey items measuring the specific individual characteristics stated in

Proposition 3. First, to determine proactive personality (M = 5.17, SD = 0.82; coefficient

α = 0.83), participants responded to a well-established reflective 10-item scale with a 7-

point Likert-type format (Seibert, Crant, & Kraimer, 1999) shown in Table 3. Second, we

measured an individual’s risk attitude (M = 6.57, SD = 2.32) by means of the widely-used

single item measure from (Dohmen et al., 2011) which was identified as the “the best all-

round predictor of risky behavior” (p. 522). It requires participants to rate their

willingness to take risks, in general, on an 11-point rating scale. Third, control appraisal

(M = 2.32, SD = 1.03; coefficient α = 0.75) was assessed with four reflective items on a

rating-scale ranging from 1 (not at all true) to 7 (very true), adapted from S. K. Parker et

al. (2006) (see Table 3). Finally, the respondents reported their experience (M = 14.49,

SD = 12.70) in the field of supply chain management (measured in years). A summary of

the measures’ descriptive statistics and bivariate correlations is shown in Table 4.

The psychometric properties of the two reflective multi-item scales (proactive

personality and control appraisal) were assessed using covariance-based confirmatory

factor analysis (CFA). The resulting fit of the measurement model to the data was

acceptable (Hair, Black, Babin, Anderson, & Tatham, 2009): χ2(53) = 83.18 with p < 0.01

(χ2/df = 1.57), CFI = 0.92, TLI = 0.91, SRMR = 0.06, and RMSEA = 0.07 (90%

confidence interval CI = [0.04, 0.10]). As shown in Table 3, the factor loadings of all

items were statistically significant (p < 0.001) and composite reliabilities (0.84 for

proactive personality and 0.76 for control appraisal) exceeded the cut-off value of 0.70

(Hair et al., 2009). The average variance extracted (AVE) values were slightly below

(0.37 for proactive personality) and above (0.52 for control appraisal) the common

threshold of 0.50.


Table 3: Multi-item measurement scales

Measures and associated indicators Coefficient

alpha

Composite

reliability λa SE

Proactive personality (Seibert et al., 1999) 0.83 0.84

Please indicate to which extent you agree with the following statements (1: strongly disagree – 7:

strongly agree)

PP1b I am constantly on the lookout for new ways to improve my life. - -

PP2 Wherever I have been, I have been a powerful force for constructive change. 0.81 0.05

PP3 Nothing is more exciting than seeing my ideas turn into reality. 0.64 0.06

PP4 If I see something I don’t like, I fix it. 0.53 0.07

PP5 No matter what the odds, if I believe in something I will make it happen. 0.52 0.08

PP6 I love being a champion for my ideas, even against others’ opposition. 0.67 0.06

PP7 I excel at identifying opportunities. 0.56 0.07

PP8 I am always looking for better ways to do things. 0.47 0.08

PP9 If I believe in an idea, no obstacle will prevent me from making it happen. 0.63 0.07

PP10 I can spot a good opportunity long before others can. 0.55 0.07

Control appraisal (S. K. Parker et al., 2006) 0.75 0.76

Please indicate to which extent you consider the following statements true (1: not at all true – 7: very

true)

CA1b In my job, most of the problems that I experience are completely “out of my

hands.” - -

CA2 With many of the problems I experience, it is not worth telling anybody

because nothing will change. 0.60 0.08

CA3 I feel powerless to control the outcomes of the process I work on. 0.91 0.08

CA4 The same problems keep happening again and again, regardless of what I do. 0.62 0.08

Note. λ refers to standardized factor loading and SE to standard error (asymptotically robust estimate). a All factor loadings are significant at the p < 0.001 level (two-tailed). b Item was excluded to increase internal consistency.

Table 4: Bivariate correlations and descriptive statistics

(1) (2) (3) (4)

(1) Proactive personality 0.37 0.12 0.02 0.00

(2) Risk attitude 0.35 *** – 0.00 0.00

(3) Control appraisal –0.13 –0.03 0.52 0.04

(4) Experience (in years) –0.07 0.02 –0.19 * –

Mean (M) 5.17 6.57 2.32 14.49

Standard deviation (SD) 0.82 2.32 1.03 12.70

Note. n = 121. Pearson product-moment correlation coefficients are shown below the diagonal, diagonal

values represent average variances extracted (where appropriate), and squared correlations (shared

variance) are above the diagonal in italics.

* p < 0.05 (equals |r| > 0.18), ** p < 0.01 (equals |r| > 0.23), *** p < 0.001 (equals |r| > 0.30) (two-tailed).

Based on the composite reliability value of proactive personality, it can be

concluded that convergent validity of the construct is adequate even though AVE is below

0.50. Moreover, since both constructs extract more variance than they share (Pearson

correlation coefficient r = –0.13; r2 = 0.02), discriminant validity is supported (Fornell &


Larcker, 1981). Having established the validity and reliability of the reflective scales, we

used scale averages as latent variable scores for the following analyses.

Consistency checks. A consistency check choice set, where one of the two proposed

proactive actions was a dominant option (all attribute levels were more desirable), was

used to assess participants’ understanding of the DCE task. Consistent participants chose

either the dominant option or the “neither” alternative. Only two of the remaining 123

respondents (1.5 %) failed this consistency check. Data from these respondents were

excluded, reducing the sample to 1089 observations (121 included participants × 9 choice

sets) for further analyses.

Before analyzing the respondents’ choice behavior, the manipulations of

vulnerability and severity in the description of the situational context were verified. To

this end, all respondents were required to answer two manipulation check questions (7-

point rating scales anchored at 1 := “not at all” and 7 := “completely”) after reading the

scenario description. To validate the manipulation of vulnerability, participants had to

evaluate whether the vulnerability to the potential disruption of the protagonist’s firm was

high. The participants’ responses were significantly different for low and high levels of

vulnerability (Mlow = 3.02, Mhigh = 5.80; t(121), p < 0.001). Perceived severity was

assessed by asking the participants to which extent they agree that the negative

consequences of the depicted supply disruption would be severe for the firm of the

protagonist. The average response of 5.93 (SD = 1.79) indicated that the participants were

well aware of the potential severe damage that a future disruption could cause.

2.4 Results

2.4.1 Estimation strategy

Discrete choice analysis uses random utility theory to provide insights into the choice

preferences of individuals (Thurstone, 1927). The main premise of random utility theory

is that a decision maker’s utility for a certain response option is determined by an

explainable systematic component and an unexplainable random component. The former

comprises observed attributes of different choice alternatives and individual

characteristics of a decision maker (Ben-Akiva & Lerman, 1985; McFadden, 1986), while

the latter accounts for all unidentified factors of a decision task (Louviere, Flynn, &

Carson, 2010). The characteristics of a decision maker are constant for each individual;


hence, they are typically considered as interaction terms with attributes or alternative-

specific constants in estimation models (Ryan, Gerard, & Amaya-Amaya, 2007).

The most widely applied model to analyze and statistically test data from DCEs is

the multinomial logit (MNL) model (also known as conditional logit model). MNL relies

on the assumption that the random errors in the utility functions of individuals are

independent and identically distributed according to Gumbel distribution. Formally,

utility is defined as:

𝑈𝑖𝑛 = 𝑉𝑖𝑛 + 𝜀𝑖𝑛. (1)

Uin is the utility of individual n for choice alternative i, Vin is the explainable

component, and 𝜀𝑖𝑛 is the unexplainable random component.

DCEs typically expose participants to choice situations with at least two response

options. The alternatives contained in each choice set are constructed by means of a set

of attributes. MNL determines the probability of selecting a specific alternative from a

set of multiple alternatives as follows (Ben-Akiva & Lerman, 1985; Louviere &

Woodworth, 1983; McFadden, 1986):

𝑃𝑖𝑗 =𝑒𝑉𝑖𝑗

∑ 𝑒𝑉𝑘𝑗𝐾𝑘=1

(2)

where Pij is the probability of choosing alternative i from choice set j out of a total

number of K possible alternatives. Vij is the explainable part of an individual’s utility

function for alternative i in choice set j. This systematic utility component can be

expressed as a function of attributes and characteristics of individual decision makers

(Lancsar & Louviere, 2008):

𝑉𝑖𝑗 = 𝛽𝑋𝑖𝑗𝑙′ + 𝛾𝑍𝑖

′ (3)

with 𝑋𝑖𝑗𝑙′ being the vector of attributes and their specific levels l of alternative j for

individual i and 𝑍𝑖′ the vector of an individual’s characteristics. β and γ are the coefficient

vectors to be estimated, typically by means of maximum likelihood estimation (Verma &

Pullman, 1998).

To analyze our DCE, we distinguish between two proactive actions (proactive

action A and proactive action B) and a “no choice” option. Accordingly, we use the

following specification of VA, VB, and Vno as the probability of choosing proactive action

A, proactive action B, or “no choice”:


𝑉𝐴 = 𝛽𝐴𝑆𝐶 + 𝛽𝑅𝐸𝐹𝐹𝑅𝐸𝐹𝐹 + 𝛽𝑆𝐸𝐹𝐹𝑆𝐸𝐹𝐹 + 𝛽𝐷𝐼𝐶𝑂𝐷𝐼𝐶𝑂 + 𝛽𝑅𝐸𝐶𝑂𝑅𝐸𝐶𝑂

+ 𝛽𝐷𝐼𝐶𝑂×𝑉𝑈𝐿𝑁𝐷𝐼𝐶𝑂 × 𝑉𝑈𝐿𝑁 + 𝛽𝑅𝐸𝐶𝑂×𝑉𝑈𝐿𝑁𝑅𝐸𝐶𝑂 × 𝑉𝑈𝐿𝑁

(4)

𝑉𝐵 = 𝛽𝑅𝐸𝐹𝐹𝑅𝐸𝐹𝐹 + 𝛽𝑆𝐸𝐹𝐹𝑆𝐸𝐹𝐹 + 𝛽𝐷𝐼𝐶𝑂𝐷𝐼𝐶𝑂 + 𝛽𝑅𝐸𝐶𝑂𝑅𝐸𝐶𝑂

+ 𝛽𝐷𝐼𝐶𝑂×𝑉𝑈𝐿𝑁𝐷𝐼𝐶𝑂 × 𝑉𝑈𝐿𝑁 + 𝛽𝑅𝐸𝐶𝑂×𝑉𝑈𝐿𝑁𝑅𝐸𝐶𝑂 × 𝑉𝑈𝐿𝑁

(5)

𝑉𝑛𝑜 = 𝛽𝑛𝑜𝑐ℎ𝑜𝑖𝑐𝑒 (6)

where βASC captures alternative-specific effects of proactive action A compared to

B, and βREFF, βSEFF, βDICO, as well as βRECO are the coefficients for response efficacy

(REFF), self-efficacy (SEFF), direct implementation costs (DICO), and relationship costs

(RECO). βDICO×VULN and βRECO×VULN are the coefficients of the suggested interactions

between response costs (DICO and RECO) and the scenario variable vulnerability

(VULN). 𝛽𝑛𝑜𝑐ℎ𝑜𝑖𝑐𝑒 reflects the utility associated with the “no choice” option. To

investigate Propositions 3a-3d, we added 16 interaction terms between all proactive

action attributes (REFF, SEFF, DICO, and RECO) and individual-specific factors,

namely proactive personality, risk attitude, control appraisal, and experience, to VA and

VB. For the sake of clarity and due to space constraints, we do not show the augmented

equations.

If a “no choice” is offered to participants in a DCE, it has been recommended to

consider the use of nested logit (NL) models (Ryan & Skåtun, 2004). To assess whether

a nested logit outperforms the MNL model formulation, we constructed a NL model that

we compared with the MNL model as delineated above. The NL model comprised the

“no choice” option as one (degenerate) nest and the two proactive response options as a

second nest. A likelihood ratio test revealed that a nested structure does not significantly

improve model fit (χ2(9) = 7.43; p = 0.59). Hence, the MNL model specification was

chosen.

2.4.2 Model estimation

Table 5 shows the results of the estimated MNL models as specified in equations 4, 5,

and 6. Model 1 serves as a baseline model, which contains only four proactive action

attributes that were incorporated in the discrete choice task and the interactions of

response cost and vulnerability, as described in Proposition 2. In a second step, we

incorporated the individual-specific factors assumed to influence choice behavior as

proposed in Propositions 3a-3d into a second model (Model 2). Log likelihood ratio tests


support the statistical significance of Model 1 (χ2(6) = 230.75; p < 0.001) and Model 2

(χ2(22) = 275.92; p < 0.001) and a likelihood ratio test between both models reveals

significant improvements from Model 1 to Model 2 (χ2(16) = 45.18; p < 0.001). Hence,

detailed results of Model 2 are discussed below.

Table 5: Estimated MNL models

Model 1 Model 2

Variable Prop. β SE p-value β SE p-value

Utility from specific proactive action attributes

Alternative-specific constant 0.16 0.11 0.13 0.18 0.11 0.11

Response efficacy

P1

1.36 0.12 0.00 *** 1.38 0.12 0.00 ***

Self-efficacy 1.55 0.14 0.00 *** 1.60 0.14 0.00 ***

Direct implementation costs –1.93 0.35 0.00 *** –1.94 0.35 0.00 ***

Relationship costs –1.79 0.29 0.00 *** –1.71 0.30 0.00 ***

Direct implementation costs × Vulnerability P2

0.48 0.19 0.01 * 0.45 0.20 0.02 *

Relationship costs × Vulnerability 0.48 0.16 0.00 ** 0.39 0.16 0.02 *

Response efficacy × Proactive personality

P3

–0.33 0.12 0.01 **

Self-efficacy × Proactive personality –0.15 0.13 0.25

Direct implementation costs × Proactive personality 0.29 0.14 0.05 *

Relationship costs × Proactive personality 0.09 0.14 0.53

Response efficacy × Risk attitude 0.20 0.12 0.09 †

Self-efficacy × Risk attitude 0.01 0.13 0.92

Direct implementation costs × Risk attitude –0.05 0.14 0.72

Relationship costs × Risk attitude 0.14 0.14 0.31

Response efficacy × Control appraisal –0.10 0.11 0.36

Self-efficacy × Control appraisal –0.09 0.12 0.47

Direct implementation costs × Control appraisal 0.26 0.13 0.06 †

Relationship costs × Control appraisal 0.33 0.13 0.01 **

Response efficacy × Experience –0.20 0.11 0.07 †

Self-efficacy × Experience –0.05 0.13 0.71

Direct implementation costs × Experience –0.14 0.14 0.32

Relationship costs × Experience 0.14 0.13 0.30

Utility from not taking proactive action

Constant 1.31 0.13 0.00 *** 1.33 0.13 0.00 ***

Log likelihood –998.09 *** –975.50 ***

Akaike information criterion (AIC) 2012.20 1999.00

McFadden’s Pseudo R2 0.10 0.12

Note. Prop. refers to proposition, β to estimated coefficients, and SE to standard error. Both models were

estimated in NLOGIT 6 using full information maximum likelihood estimators based on 1089

observations (121 participants × 9 choice sets). The variables proactive personality, risk attitude, control

appraisal, and experience have been standardized to facilitate the interpretation of the estimated effects.

† p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed).

As a robustness check, a further MNL model was estimated that specifically

controlled for effects of the selected language (English/ German) on the relative


importance of proactive action attributes. Although results slightly changed

quantitatively, interpretations remained qualitatively similar to those reported in Table 5.

Coping Appraisal Variables. All the attributes used to construct proactive response

options (response efficacy, self-efficacy, direct implementation costs, and relationship

costs) showed a statistically significant effect on the decision between different types of

proactive actions. This lends empirical support for Propositions 1a, 1b, and 1c. As

suggested by PMT, response efficacy (βREFF = 1.38, p < 0.001) and self-efficacy (βSEFF =

1.60, p < 0.001) increase the probability of selecting a specific alternative, whereas direct

implementation costs (βDICO = –1.94, p < 0.001) and relationship costs (βRECO = –1.71, p

< 0.001) reduce the latter. The constant term in equation 4 captured alternative-specific

effects of proactive action A compared to proactive action B and did not show a

statistically significant effect on choice behavior (βASC = 0.18, p = 0.11) which strengthens

the validity of the chosen experimental design, because a generic label was used for both

of these response alternatives. The relative influence of specific components of the

employed model is depicted in Figure 3. Following Verma et al. (2006), we set the most

influential β-coefficient (i.e., direct implementation costs) equal to 1 and rescaled all

remaining coefficients relative to it between 0 and 1.

Figure 3: Relative importance of DCE response alternative variables

Note. The highest coefficient (direct implementation costs) is set to 1. All other coefficients are rescaled

accordingly.

Individual Characteristics. The MNL also allowed for an analysis of further

components that are assumed to affect an individual’s intention to take proactive action.

All four included decision maker characteristics revealed (marginally) significant

interaction effects with the relative importance of certain proactive action attributes,

providing support for Propositions 3a-3d. The higher an individual’s proactive

personality, the lower the importance of response efficacy (βREFF×PP = –0.33, p = 0.01)


and direct implementation costs (βDICO×PP = 0.29, p = 0.04) for the decision whether to

engage in proactive action. Moreover, for more risk-seeking participants, the relative

importance of response efficacy increased (βREFF×RISK = 0.20, p = 0.09). Finally, control

appraisal reduced the impact of response costs (βDICO×CA = 0.26, p = 0.06; βRECO×CA =

0.33, p = 0.01) on the choice of proactive action alternatives and higher experience

resulted in a lower relative importance of response efficacy (βREFF×EXP = –0.20, p = 0.07).

In addition, after participating in the DCE, the respondents were asked to directly

rate the perceived relative importance of the four selected proactive action attributes. The

results revealed that, in contrast to results of the DCE, the most influential attribute was

perceived to be response efficacy (M = 6.33, SD = 0.77) followed by self-efficacy (M =

5.21, SD = 1.32). Direct implementation costs (M = 4.94, SD = 1.34) and relationship

costs (M = 4.52, SD = 1.46) were perceived as less important, as shown in Figure 4.5

Figure 4: Perceived relative importance of DCE response alternative variables

Vulnerability and Response Costs. As suggested in the second proposition, in

addition to the main effects of the coping appraisal variables, vulnerability affected the

relative importance of response costs in the selection of alternative proactive actions.

When vulnerability is high, direct implementation costs (βDICO×VULN = 0.45, p = 0.02) and

relationship costs (βRECO×VULN = 0.39, p = 0.02) of proactive response options are less

important for the choice of a specific proactive action.

5 Pairwise comparisons revealed that, except for the mean difference between the relative importance of

self-efficacy and direct implementation costs (p = 0.14), all mean differences were statistically significant

(p < 0.05).


2.5 Discussion

2.5.1 Theoretical implications

This study has several important theoretical implications for decision making in the

context of supply risks and disruptions. First, PMT has primarily been applied to health-

related threats that directly concern an individual. This research supports the idea that

PMT is more widely applicable and an insightful framework for situations involving

almost any kind of threat. Instead of using a threat to individuals that would involve direct

emotional or physical harm, we applied PMT to investigate choice behavior of individuals

facing a threat to an organization. As per our propositions, the DCE revealed that trade-

offs are made in supply risk management to select the most attractive proactive measures.

All main variables of PMT’s coping appraisal showed statistically significant effects on

the participants’ choice behavior. The estimated utility that individuals obtain from

evaluating the supply disruption scenario and potential proactive measures is an indicator

of their protection motivation.

As shown in Figure 3, the most important variables when selecting proactive

measures in supply risk management are the two types of response costs variables. High

direct implementation or relationship costs can render a proactive measure so unappealing

that even high response efficacy or high self-efficacy alone cannot offset this burden. The

DCE results, nevertheless, emphasize that self-efficacy is a crucial component that affects

the behavior of individuals within organizations, as delineated by prior research

(Bandura, 1977). Most surprisingly, the perceived effectiveness of a response alternative

in mitigating future loss, which is the main aim of implementing proactive measures, has

the lowest relative importance among the included attributes. Finally, although “doing

nothing” might appear socially undesirable at first sight, the “no choice” option was the

preferred choice in 51.2% of all 1089 choice sets in the final sample.

Second, the role of individual characteristics in the DCE and subsequent analyses

demonstrate that there is a need to account for the role of individuals within decision

making processes in supply chain risk management. Figure 5 shows an enriched version

of PMT, which includes an additional layer depicting the identified interactions of coping

appraisals with individual-specific characteristics as a more comprehensive model of

proactive behavior.


Proactive personality, risk attitude, control appraisal, and experience had

statistically significant effects on the relative importance of certain proactive action

attributes. Moreover, the results of this study imply that the components of PMT differ

regarding their relative importance. Figure 3 shows that a surprising result of the DCE is

that both types of response costs emerged as more decisive variables in determining

whether to proactively mitigate future losses than, for instance, the response efficacy of

a specific action.

Figure 5: Enriched model of protection motivation theory

Third, the insights generated reveal that the perception of decision makers about the

relative importance of certain attributes of a proactive action deviates considerably from

their actual choice behavior. For instance, although our participants perceived a proactive

action’s effectiveness as most decisive for their choice, they subconsciously assigned

considerably less importance to it when actually selecting an action. This is an important

issue, which could be the result of an overly strong focus on costs that supply chain

professionals are not aware of. Our study is, to the best of our knowledge, among the first

efforts to empirically identify mismatches between perceptions and actual behavior of

professionals in the context of procurement and underlines the need for decision makers

to better understand their own choice behavior.

Fourth, the DCE data revealed that the characteristics of a threat affect the relative

importance of specific attributes of response alternatives. The effect of direct

implementation costs on the behavior of individuals depends on the vulnerability to a

threat. As can be inferred from Figure 3, direct implementation costs have a smaller

impact on the selection of specific proactive measures in scenarios where the vulnerability

Response efficacy

Self-efficacy

Sources of

information Cognitive mediating processes Coping modes

Environmental

- Verbal

persuasion

- Observational

learning

Intrapersonal

- Personality

variables

- Prior experience

Action or inhibition

of action

- Single act

- Repeated acts

- Multiple acts

- Repeated,

multiple acts

Factors affecting response probability

Maladaptive

response

Adaptive

response

Increasing Decreasing

Intrinsic rewards

Extrinsic rewards

Severity

Vulnerability

Response costs

=

=

Protection

motivation

Fear

arousal

Threat

appraisal

Coping

appraisal

Proactive personality

Risk attitude

Control appraisal

Experience

+

++Individual-

specific

factors


to a threat is high rather than low. This is intuitive in the sense that greater vulnerability

might lead to a greater tolerance for higher response costs because, in this case, a threat

arising from a supply disruption could appear more unavoidable.

2.5.2 Managerial implications

In addition to the delineated theoretical implications, this study’s findings entail

considerable implications for managerial practice. One implication of this study is the

way in which individuals process insights from observing competitors exposed to supply

disruptions to adjust their own management of supply risk. Prior research has already

suggested that vicarious learning is a relevant issue in the management of supply risk and

disruptions (Hora & Klassen, 2013). We contribute to these insights by adding new details

on how exactly information on someone else’s misfortune and an impending disruption

can be translated into proactive action.

Individual managers focus on the costs of a specific action when they decide

whether or not to act proactively, although they perceive self-efficacy and the action’s

effectiveness as more important. This mismatch between perceptions and actual behavior

underlines the need for decision makers to improve their understanding of how they make

choices. Moreover, although the DCE focused on the example of a severe disruption that

might lead to tremendous (financial) loss, the costs of a measure instead of its ability to

mitigate future loss were extremely decisive for individual decision makers. This might

reveal a tendency to admit too much relevance to the costs of a response in the selection

process of proactive measures which might not always be desirable. The estimated loss

associated with a supply disruption is often difficult to discern; this makes it challenging

to weigh the costs of a proactive measure against the potential damage.

Finally, in supply risk management, firms depend to a large part on the self-efficacy

of their employees (as seen in Figures 3 and 4), which is even more important than the

perceived effectiveness of a response option in avoiding the dangers of a threat. Hence,

firms benefit from being aware of this circumstance and training their employees to

appropriately assess their own abilities and build confidence. Otherwise, a manager who,

mistakenly, does not feel capable of successfully performing a certain proactive action

will avoid selecting it, which might result in an unnecessarily high exposure to future

damage through supply disruptions for a firm. In addition, the results depict several


effects of individual characteristics on proactive behavior. Thus, these insights might be

helpful in recruiting decisions and personnel allocation.

2.5.3 Limitations and future research opportunities

Several limitations constrain the contribution of this study, but, at the same time, also

highlight fruitful avenues for future research. First and foremost, the choice experiment

relies on stated intentions of individuals instead of their real-life behavior. Prior research

indicates that intentions can be reliable indicators of actual behavior (Ajzen, 1991; T. L.

Webb & Sheeran, 2006). However, in the event of an actual impending supply disruption,

individuals might behave differently than indicated in response to the hypothetical choice

situation, such as environmental factors or time pressures unaccounted for in our DCE.

Furthermore, the experimental task is limited to decision makers with centralized

authority because we assume that this is a basic characteristic of risk management

processes. This means that, at the same time, we did not consider many factors that might

also influence the selection process of proactive measures, such as the presence of a

hierarchy (Mihm, Loch, Wilkinson, & Huberman, 2010), the need to coordinate with

others (Lounamaa & March, 1987), or adding other characteristics of an individual

manager’s work environment (S. K. Parker et al., 2006).

Following previous research, we refrained from explicitly distinguishing response

costs and probable extrinsic as well as intrinsic rewards from maladaptive behavior.

However, since these rewards are part of the PMT’s threat appraisal process, the

consideration and implementation of specific rewards of, for example, not taking

proactive action within an experimental design might provide an even more sophisticated

picture of proactive risk management decisions. In addition, we used four two-level key

variables of PMT as attributes to construct response options. As this was the first attempt

to model the selection process of proactive actions in supply risk management using

discrete choice modelling, this study has an exploratory character and adding attributes

or distinguishing among further levels to describe response alternatives in the DCE seems

a promising way to generate additional insights.

Another limitation is that choice behavior might vary with the cultural background

of respondents as demonstrated by S. C. Schneider and De Meyer (1991). They showed

that Latin European managers were more likely to respond proactively to strategic issues

than their North American, British, northern European, or Nordic counterparts. Hence, to


make the generated insights more generalizable, we encourage the validation of our

results by means of a larger sample comprising supply chain professionals from several

cultural regions.

The identification of the main drivers of proactive behavior in supply risk

management provides insights into decision making behavior. The results of this study

show that PMT is an insightful framework to analyze the selection of proactive measures.

Moreover, the identified mismatch between perceptions and actual choice behavior is an

interesting topic for future research. The designed DCE can serve as a starting point for

future research to further improve our understanding of how managers cope with the

threat of supply disruptions and for practitioners to develop training tools for proactive

decision making in supply risk management.

Finally, this research focused on supply disruptions that are accompanied by a

potentially severe negative impact on the performance of a firm and cannot be solved

within day-to-day operations. Choice behavior and the underlying relative importance of

specific attributes might considerably diverge between exposure to less severe instead of

very severe threats, as already suggested by prior research (Cismaru & Lavack, 2007). It

is important to understand the performance implications of these choices when the threat

is minor and severe.

Substantive and symbolic corporate social responsibility: Blessing or curse in case of misconduct? 34

Chapter 3 Substantive and symbolic corporate

social responsibility: Blessing or curse in

case of misconduct?

Co-author:

Christoph Bode


Abstract

Often, corporate social responsibility (CSR) is associated solely with “doing good,” but

firms also have to prevent corporate social irresponsibility (CSIR) (i.e., “avoiding bad”).

Many firms engage in CSR – either substantively or only symbolically – in the hopes that

a reputation for CSR mitigates negative stakeholder reactions in case the firms suddenly

become involved in a CSIR incident. Yet, research on the effects of a CSR reputation on

stakeholder reactions to CSIR is equivocal. Some studies have theorized insurance-like

effects of ex ante CSR, whereas others have suggested the exact opposite, namely that a

reputation for CSR may even aggravate negative reactions to CSIR. Moreover, extant

research on stakeholder reactions to CSIR has focused chiefly on consumers and

investors, although stakeholders increasingly hold firms accountable for misconduct

within their supply chains, which has repercussions on supplier selection decisions. The

present study is innovative in that it focuses on the business-to-business (B2B) context

from a purchasing perspective and proposes a model that explains the conditions under

which CSR acts as an insurance or a liability subsequent to CSIR. The empirical results

from a vignette experiment with supply chain managers add to the understanding of the

effects of CSR activities on negative stakeholder reactions to CSIR and provide important

theoretical and practical implications.


3.1 Introduction

The issue of corporate social irresponsibility (CSIR) has only rarely been addressed in

the corporate social responsibility (CSR) literature, although stakeholders devote growing

attention to environmental and social misconduct (Fiaschi et al., 2017). Stakeholders exert

increased pressure on firms to behave in socially responsible ways (Campbell, 2007) and

increasingly hold them responsible for misconduct in their supply chains (Hartmann &

Moeller, 2014; Y. H. Kim & Davis, 2016). The academic literature has strongly focused

on linking CSR with the concept of “doing good,” while “avoiding bad” also constitutes

an important condition for firms to be perceived as socially responsible (Lin-Hi & Müller,

2013). CSIR applies to all industries and it can take various forms from environmental

disasters (e.g., the Deepwater Horizon oil spill in 2010) and workplace disasters (e.g., the

Rana Plaza building collapse in 2013) all the way to corruption and collusion scandals

(e.g., the price-fixing cartel of the truck makers DAF, Daimler, Iveco, and Volvo-Renault

between 1997 and 2011). The globalization of markets and increasing interconnectedness

of supply networks have recently added to the complexity that firms face and made it

even more challenging for managers to prevent CSIR. In the light of this development, it

is important to understand that consumers not only blame firms for CSIR that occurs

inside their own barriers but also hold buying firms responsible for their suppliers’

environmental and social misconduct (Hartmann & Moeller, 2014). Subsequent negative

stakeholder reactions pose substantial risks that need to be addressed (Lin-Hi &

Blumberg, 2018). Since firms are perceived to be only as responsible as their supply

network (Andersen & Skjoett-Larsen, 2009), the purchasing function plays a key role in

preventing CSIR and addressing these risks. This study focusses on the role of the

purchasing function as gatekeeper to CSIR in the supply chain of the focal firm.

Research has been concerned with whether a firm’s ex ante CSR activities affect

negative stakeholder reactions to CSIR. Most studies have focused on consumer or

investor reactions and theorized an insurance-like mechanism of a firm’s reputation for

CSR, which builds a “reservoir of goodwill” among the firm’s stakeholders and mitigates

negative responses to bad news (e.g., Flammer, 2013; Godfrey et al., 2009). But there are

also studies purporting the exact opposite, namely that firms engaging in CSR experience

more negative reactions to CSIR incidents than firms that do not promote themselves as

socially responsible (e.g., Sen & Bhattacharya, 2001; Swaen & Vanhamme, 2003;

Vanhamme & Grobben, 2009). Indeed, recent examples, such as the German car


manufacturer Volkswagen (VW) that won numerous CSR awards but currently faces

severe public criticism due to the “Dieselgate” scandal, cast doubts on insurance-like

effects of ex ante CSR (Lynn, 2015).

Like brand commitment or consumer-company identification, a CSR reputation

might serve as a goodwill buffer in case of non-severe CSIR, but also an expectation

burden in case of severe CSIR (e.g., Einwiller, Fedorikhin, Johnson, & Kamins, 2006;

Germann et al., 2014; Janssen et al., 2015; Kang et al., 2016). Based on assimilation-

contrast theory, this research contributes to identifying the boundary conditions of the

“insurance effect” of a CSR reputation thereby addressing recent calls for more research

on how ex ante CSR affects negative stakeholder reactions in the aftermath of CSIR (e.g.,

Kang et al., 2016; S. Kim & Choi, 2016; Lenz, Wetzel, & Hammerschmidt, 2017). More

specifically, this study investigates how negative stakeholder reactions to CSIR in

industrial buying contexts are shaped by the type of CSR reputation and the CSIR

severity. To this end, the employed research design (randomized vignette experiment)

accounts for different approaches to build a CSR reputation (substantive vs. symbolic).

The empirical results suggest that CSR mitigates negative reactions to non-severe CSIR

but that CSR reputations driven by symbolic actions aggravate negative stakeholder

reactions in case of severe CSIR. In addition, the results provide important and innovative

insights for managers who are concerned with stakeholder management and the allocation

of resources to CSR activities while facing the risk of CSIR.

3.2 Conceptual background and hypotheses development

3.2.1 Substantive and symbolic management of corporate social responsibility

expectations

Although the literature on CSR is vast, there are still controversies revolving around the

definition of CSR (e.g., Colombo, Guerci, & Miandar, 2017; Sheehy, 2015). As Lin-Hi

and Müller (2013) stated, “despite the growing interest in this topic, there is still no

general agreement on the precise meaning of CSR” (p. 1928). CSR has often been used

as an umbrella term for concepts, such as sustainability, business ethics, or corporate

citizenship (de Jong & van der Meer, 2017; Freeman & Hasnaoui, 2011). Nevertheless,

the accepted idea of CSR is that society and business are not autonomous but

interdependent and that “society has certain expectations for appropriate business

behavior and outcomes” (Wood, 1991, p. 695). For the purpose of this study, following


McWilliams and Siegel (2001), we define CSR as a firm’s actions that appear to “advance

some social good, beyond the interests of the firm and what is required by the law” (p.

117). Moreover, we adopt a holistic, multidimensional view of CSR. Typically,

environmental, social, and governance (ESG) activities are captured to determine the

degree to which a firm lives up to its CSR (Arvidsson, 2010; Cheng, Ioannou, & Serafeim,

2014). Examples of such activities include the use of renewable raw materials (Ketola,

2010), the establishment of corporate foundations (Westhues & Einwiller, 2006), and the

presence of an external auditor to examine, verify, and validate a CSR report (Lynes &

Andrachuk, 2008). However, research concerned with CSR has often focused only on

single dimensions (e.g., Walls, Berrone, & Phan, 2012) and neglected this

multidimensionality (e.g., Bénabou & Tirole, 2010; Carroll, 1979; Waddock & Graves,

1997).

Firms may establish and maintain a reputation for CSR by engaging in CSR-related

activities (McWilliams & Siegel, 2001; Vanhamme & Grobben, 2009). More specifically,

from a stakeholder perspective, firms have to address their stakeholders’ demands, such

as customers, suppliers, employees, or local communities, who have expectations with

regard to a firm’s social responsibility. A firm can obtain support for its operations and,

from an institutional perspective, maintain its “social license to operate” (SLO)

(Demuijnck & Fasterling, 2016, p. 675) only when the behavior of a firm is in line with

these demands. In addition, buying firms are increasingly held responsible for their

suppliers’ behavior and criticized as soon as this behavior deviates from the stakeholders’

expectations (Hartmann & Moeller, 2014). This “chain liability” adds to the challenge of

maintaining an SLO, because it implies that a buying firm is only as socially responsible

as its supply network (Andersen & Skjoett-Larsen, 2009; Krause, Vachon, & Klassen,

2009). Consequently, the purchasing function plays a key role in implementing a firm’s

CSR strategy by mitigating risks associated with environmental or social misconduct of

suppliers (Hajmohammad & Vachon, 2016; L. Schneider & Wallenburg, 2012).

An SLO provides a basis for firm’s activities to be perceived as legitimate in the

eyes of stakeholders (Demuijnck & Fasterling, 2016). Legitimacy is typically defined as

“a generalized perception or assumption that the actions of an entity are desirable, proper,

or appropriate within some socially constructed system of norms, values, beliefs, and

definitions” (Suchman, 1995, p. 574). According to institutional theory, firms can use two

strategies to address their stakeholders’ demands and obtain legitimacy. They can pursue


either a substantive or a symbolic adaptation approach (Ashforth & Gibbs, 1990). From a

stakeholder management perspective, the overall aim of these approaches is to manage

stakeholders’ perceptions of meeting societal expectations by either engaging in actions

entailing real change or claims/ promises providing representations of such actions

(Wickert, Scherer, & Spence, 2016). Highhouse, Brooks, and Gregarus (2009) developed

a model showing that these activities serve as cues about a firm’s CSR policy, which are

processed by individuals to form a perception about a firm’s reputation.

The substantive adaptation approach involves considerable changes in core

procedures or long-term investments, which entail certain risks but ensure actual

compliance with the expectations imposed by the external environment (Eccles, Ioannou,

& Serafeim, 2014). Substantive actions include, for example, the use of renewable energy,

the development of products that provide specific health or safety benefits, and the

acquisition of an above-average percentage of independent board members. In contrast,

the symbolic adaptation approach is based on activities that seek to decouple the firm’s

actual practices from the external demands by means of superficial actions that merely

show “ceremonial conformity” but do not necessarily have any substance (Ashforth &

Gibbs, 1990; Meyer & Rowan, 1977). Examples are the formation of a CSR committee,

a membership in a voluntary initiative that aims to reduce CO2 emissions, and the mere

claim to provide flexible working hours to employees. These actions do not necessarily

entail real and concrete change in business processes, but they have the potential to be

utilized as a cover for poor actual CSR performance (Russo & Harrison, 2005). Symbolic

responses to stakeholder demands aim at producing “impressions of more material

change” (Durand, Hawn, & Ioannou, 2017, p. 5) and managing stakeholder perceptions

of environmental and social commitment (Bansal & Clelland, 2004). Despite the

circumstance that symbolic activities do not involve concrete changes in organizational

procedures, they may suffice to promote a firm’s legitimacy because the “appearance

rather than the fact of conformity is often presumed to be sufficient for the attainment of

legitimacy” (Oliver, 1991, p. 155). In line with this, several empirical studies have

suggested that the use of symbolic CSR actions, decoupled from concrete change,

positively affects a firm’s legitimacy (e.g., Weaver, Trevino, & Cochran, 1999; Westphal

& Zajac, 2001; Zott & Huy, 2007). However, although both substantive and symbolic

CSR engagement may translate into being perceived as legitimate, further empirical

research shows that substantive and symbolic CSR activities may differ with regard to


their implications for stakeholder attributions and behavior (e.g., Donia, Ronen, Sirsly, &

Bonaccio, 2017; Godfrey et al., 2009; McShane & Cunningham, 2012; Vlachos,

Panagopoulos, & Rapp, 2013).

It is not surprising that, all else equal, managers tend to prefer to pursue the less

time-consuming and resource-intensive stakeholder management via symbolic

assurances (Ashforth & Gibbs, 1990). This preference for symbolic assurances carries a

risk, because stakeholders typically demand substantive action. If noticed, the use of

symbols, claims, and promises without actually providing a social good could lead to a

loss of all benefits generated from previous CSR activities (Hawn & Ioannou, 2016). If

symbolic CSR actions were interpreted as “greenwashing” for the mere sake of being

granted legitimacy, firms may be perceived as untrustworthy and manipulative (Walker

& Wan, 2012). Being one of the most prominent industry examples of the last decades,

British Petroleum’s (BP) symbolic commitment to CSR before the Deepwater Horizon

catastrophe highlights that such decoupling can be enough to obtain legitimacy. However,

this example also highlights that whether substantive or symbolic activities have been

chosen to achieve legitimacy has important repercussions in case of severe misconduct

because the “beyond petroleum” campaign based on symbolic CSR engagement first

helped BP to be perceived as socially responsible, “before being turned against it as a

testament of perceived greenwashing” (Matejek & Gössling, 2014, p. 579).

3.2.2 Corporate social irresponsibility

The examination of CSIR in the academic literature started with Armstrong (1977) and

the topic has been only rarely addressed since, although “avoiding bad” is considered a

precondition for a firm to be perceived as a responsible actor (Lin-Hi & Müller, 2013). In

line with Lin-Hi and Müller (2013), we define CSIR as firm-induced incident “that results

in (potential) disadvantages and/ or harm to other actors” (p. 1932). This includes, for

instance, the release of toxic chemicals into waterways (Greenpeace, 2014), corruption

scandals (Clark, 2010), and labor law violations (Reuters, 2014). Such events are frequent

and widespread across industries, and the probability of the occurrence of CSIR is a

function of the complexity of a firm’s business (Vanessa, Jijun, & Bansal, 2006). CSIR

incidents may trigger various consequential negative stakeholder reactions such as

penalties, compensation payments, sales bans, and decreased employee motivation, but

also reputational damage and customer losses, which can even lead to the demise of firms


(e.g., Jin, 2016; Sims & Brinkmann, 2003). To manage the risk of CSIR in their supply

chains and curb the chain liability effect, firms can choose from a broad range of actions

to respond to misconduct of suppliers. Contractual agreements between a focal firm and

its suppliers (e.g., phase-out of a supplier instead of immediate termination of the

relationship) might constrain these response actions.

Many firms pursue CSR-related actions in the belief that this protects them from

future reputational damage (Janssen et al., 2015; Vanhamme, Swaen, Berens, & Janssen,

2015). Still, several industry examples of CSIR have demonstrated that the effects of a

reputation for CSR on stakeholder reactions to irresponsible behavior are more complex

than a simple “insurance mechanism” would suggest. VW was highly praised for its

strong commitment to CSR until the public was informed that in fall 2015 that the firm

cheated on the pollution tests of their diesel vehicles by using an illegally manipulating

software. This scandal has already cost VW several billions of U.S. dollars, but the

corresponding negative consequences might still not be discernible to the full extent. The

firm has been heavily criticized in public, although other car manufacturers were, and still

are, exposed to similar accusations (Mehrotra, 2018). Another example concerns the

“Deepwater Horizon” oil spill of BP. For years prior to the disaster, BP has spent many

resources on its “Beyond Petroleum” campaign to be perceived as a socially responsible

firm. Nevertheless, the firm has suffered tremendously from the disaster in 2010. It is

argued that a driver of the stakeholder criticism was the firm’s CSR engagement prior to

the event (Janssen et al., 2015).

Moreover, several published studies obtained results that the “insurance

mechanism” is not able to explain. These studies suggest that under certain conditions,

mitigating effects resulting from a CSR reputation are fragile, and ex ante CSR

engagement may even lead to more negative stakeholder reactions subsequent to CSIR

compared to firms that do not have a CSR reputation. For instance, if the domain of the

CSIR incident is related to a firm’s prior CSR activities, a CSR reputation acts as a

liability that aggravates negative reactions (Wagner, Lutz, & Weitz, 2009). The recent

study by S. Kim and Choi (2016) complements this finding, demonstrating that the effect

of post-crisis communication of CSR activities on negative stakeholder reactions depends

on the domain of pre-crisis CSR initiatives conducted by the firm facing the crisis. The

domain of a firm’s post-crisis CSR engagement is also decisive for its implications on

firm value. If the engagement is related to the domain of a firm’s CSIR, it is perceived as


insincere while it may enhance firm value if it is related to other domains (Lenz et al.,

2017). In addition, it is important to consider the channels through which firms

communicate CSR activities prior to CSIR, especially when these activities do not relate

to the domain of the CSIR incident. CSR activities that are communicated through highly

credible third-party sources augment negative stakeholder reactions while CSR

information that is communicated using firm-controlled sources is able to attenuate

adverse responses compared to firms that do not at all communicate their CSR activities

(Vanhamme et al., 2015). Finally, the results of an experimental study revealed that firms

with a brief CSR history may experience more negative stakeholder reactions to corporate

crises when they use their CSR engagement in their post-CSIR communication, whereas

firms with a long CSR history can benefit from mentioning their involvement

(Vanhamme & Grobben, 2009). All these research efforts highlight that insurance-like

effects from ex ante CSR cannot be taken for granted.

3.2.3 Assimilation and contrast effects subsequent to CSIR

Previous research has strongly focused on theorizing an insurance-like mechanism of a

firm’s reputation for CSR in case of misconduct, however, the boundary conditions of

insurance-like effects of ex ante CSR on negative stakeholder reactions to CSIR remain

unclear. CSR-related activities do not necessarily need to translate into positive effects

for a firm, especially if this involvement is perceived as insincere (Sen & Bhattacharya,

2001). When a firm with a reputation for CSR is involved in CSIR, customers can lose

their more positive perception and trust this firm less than if it would have not been

promoted as socially responsible (Swaen & Vanhamme, 2003).

Assimilation-contrast theory describes how individuals evaluate new information.

Its underlying idea is that an individual’s initial expectations towards an issue serve as a

reference point to which the new information is compared (Hovland, Harvey, & Sherif,

1957; Sherif & Hovland, 1961). Depending on the extent to which the new information

violates the initial expectations, the disparity will either be assimilated towards the

reference point or contrasted away from it. In line with this, Kang et al. (2016) argued

that in a fashion similar to that of brand commitment in case of product recalls (Germann

et al., 2014) or consumer-company identification in times of negative publicity (Einwiller

et al., 2006), a reputation for CSR might attenuate negative stakeholders’ reactions to

light, non-severe CSIR but augment negative responses to severe CSIR. Put differently,


stakeholders might respond more negatively to severe CSIR than if they would have not

perceived this firm as socially responsible. Based on insights from research on

assimilation-contrast theory (e.g., R. E. Anderson, 1973), the framework for

understanding the roles of CSR in crisis situations developed by Janssen et al. (2015)

suggests that assimilation and contrast effects are not only driven by the characteristics

of CSIR (e.g., its severity) but also by stakeholder perceptions of CSR motives. These

motives are either extrinsic (akin to symbolic CSR) or intrinsic (akin to substantive CSR).

Intrinsically motivated CSR engagement is perceived as acting out of profit-driven self-

interest while extrinsically motivated CSR activities are interpreted as genuine concern

for environmental and social concerns (Batson, 1998). Assimilation-contrast theory

provides valuable insight into how reputations for CSR affect negative stakeholder

reactions to CSIR and, more specifically, into the boundary conditions of the insurance-

like mechanism of ex ante CSR.

To operationalize the reactions of customers to CSIR, we focus on two important

customer outcomes of buying situations, purchase intention (PI) and the intention to

engage in negative word-of-mouth (nWOM) (Maxham & Netemeyer, 2003). They cover

two facets of customer behavior, and they are well-established and validated in the B2B

marketing literature (Leroi-Werelds, Streukens, Brady, & Swinnen, 2014). Purchase

intentions depict how a customer intends to act in a specific buying situation while

intentions to engage in nWOM represent customers’ behavior after the buying situation.

Two unique characteristics distinguish nWOM from purchase intentions, the effect of

nWOM tends to last longer and nWOM is a potential source of information (Ham & Kim,

2017). Hence, nWOM can have tremendous consequences and may affect the purchase

decisions of not only a supplier’s existing, but also potential customers (Ferguson &

Johnston, 2011; Money, Gilly, & Graham, 1998). In industrial buying contexts, nWOM

could be shared through, for instance, supplier-selected referrals (Hada, Grewal, & Lilien,

2014). Industrial buyers tend to rely on referrals even more than consumers (Wangenheim

& Bayón, 2007) and considerable empirical evidence shows that both purchase intention

and nWOM are related to actual behavior (E. W. Anderson, Fornell, & Lehmann, 1994;

Morgan & Rego, 2006). A joint focus on both variables captures a thorough picture of

the behavior of professional buyers.

As an important characteristic of CSIR, the severity of environmental or social

misconduct might moderate the link between CSIR and negative stakeholder reactions.


CSIR of low severity could result in assimilation effects if a firm has had a reputation for

CSR prior to the incident. In this case, the disparity between a stakeholder’s expectations

and the firm’s true CSR performance might be small enough to be tolerated and

assimilated towards the more positive initial perception by stakeholders. Assimilation-

contrast theory suggests that these assimilation effects lead to more favorable stakeholder

reactions than if a firm would have not positioned itself as socially responsible. In

addition, since non-severe CSIR does not fundamentally call into question a firm’s

reputation for CSR and its corresponding motives (Janssen et al., 2015), both substantive

and symbolic CSR are expected to produce these insurance-like effects. Hence, to

investigate whether firms with reputations for substantive or symbolic CSR will

experience assimilation effects that attenuate negative stakeholder reactions in case of

non-severe CSIR, we hypothesize the following:

Hypothesis 1. If a firm with a reputation for substantive CSR is involved in non-

severe CSIR, it experiences smaller negative effects on its customers’

(a) purchase intention and

(b) intention to engage in negative word-of-mouth

than firms that do not have a reputation for CSR.

Hypothesis 2. If a firm with a reputation for symbolic CSR is involved in non-

severe CSIR, it experiences smaller negative effects on its customers’




According to assimilation-contrast theory, thresholds exist for both acceptance and

rejection and stakeholders will not accept a further increase in disparity beyond the

threshold of acceptance; therefore, severe CSIR that considerably fails to meet a

stakeholder’s initial expectations based on a firm’s ex ante CSR reputation might result

in contrast effects. We argue that this might not always be true. In the context of CSR,

the effectiveness of a firm’s actions used to be perceived as socially responsible can

depend on their nature. Previous research has highlighted that substantive and symbolic

CSR activities do not necessarily affect stakeholder reactions similarly (e.g., Donia et al.,

2017; Hawn & Ioannou, 2012). More specifically, in case of considerable violations of

prior expectations, it does matter for stakeholders whether legitimacy was achieved by

engaging mainly in symbolic rather than substantive CSR activities (Carlos & Lewis,

2018; Lyon & Maxwell, 2011; Zavyalova, Pfarrer, Reger, & Shapiro, 2012). Symbolic

CSR actions can be perceived as an ineffective attempt to meet stakeholder expectations

when the disparity between expected and true performance is large. If identified as


extrinsically-motivated greenwashing, symbolic management is likely to be punished

(Forehand & Grier, 2003; Janssen et al., 2015).

Accordingly, for severe CSIR incidents, we argue that firms which have established

a reputation for CSR driven by symbolic engagement will experience more negative

stakeholder reactions compared to firms, which have no CSR reputation (all else being

equal). The underlying logic is that the CSIR incident creates a large disparity in the

stakeholder expectations formed by a firm’s CSR reputation, which leads to contrast

effects. Hence, the customer magnifies the perceived disparity between the incident and

the initial expectations and reacts even more negatively to the incident than if the firm

would have not promoted itself as socially responsible. Prior research has suggested that

in visibly polluting industries, CSR engagement can harm corporate financial

performance (Walker & Wan, 2012). Strong and clear negative information concerning a

CSIR incident puts in doubt the credibility of ex ante CSR engagement and erodes the

overall legitimacy of the firm maneuvering a firm in a worse position than if stakeholders

had no information about its CSR activities (Yoon, Gürhan‐Canli, & Schwarz, 2006).

Firms that have established a CSR reputation based on substantive engagement are

likely to experience more favorable blame attributions about crisis responsibility since

their CSR motives are perceived to be intrinsically motivated. Extant empirical research

highlights that, even in case of severe incidents, stakeholders will be more likely to

attribute CSIR to bad luck rather than malevolence or develop alternative explanations if

stakeholders believe in intrinsic CSR motives (Godfrey et al., 2009; Minor & Morgan,

2011). Accordingly, we propose:

Hypothesis 3. If a firm with a reputation for substantive CSR is involved in severe

CSIR, it experiences smaller negative effects on its customers’




Hypothesis 4. If a firm with a reputation for symbolic CSR is involved in severe

CSIR, it experiences larger negative effects on its customers’




3.3 Method

We used a vignette-based experimental approach to test our hypotheses on the effect of a

firm’s CSR reputation in the aftermath of CSIR in B2B contexts. A vignette is typically


defined as a “short, carefully constructed description of a person, object, or situation,

representing a systematic combination of characteristics” (Atzmüller & Steiner, 2010, p.

128). Recently, vignette-experiments have been employed in the operations management

context to investigate make-or-buy decisions (Mantel, Tatikonda, & Liao, 2006),

observational learning (Hora & Klassen, 2013), and perceptual differences between

buyers and suppliers (Ro, Su, & Chen, 2016).

This approach has several methodological advantages. First, vignette-based

experiments provide a controlled test of the hypothesized causal relationships by carefully

manipulating the vignettes presented to the participants. Second, compared to

retrospective surveys or case studies, vignette-based experiments can generate more

reliable data, because the participants have to indicate their intentions shortly after reading

a specific scenario, which minimizes retrospective bias (Wathne, Biong, & Heide, 2001).

Furthermore, given the sensitive nature of environmental and social misconduct, a

vignette-based experimental design can minimize social desirability bias

(Rungtusanatham et al., 2011; Wason et al., 2002). Third, an experimental design enables

researchers to study behavior and choices where individuals or firms are usually not likely

to share information (Rungtusanatham et al., 2011). A CSIR incident typically has

adverse effects on firms. Moreover, severe incidents are rather rare, and it would be

unethical to disrupt a firm to collect data. Vignette-based experiments help overcome both

issues.

We used an online experiment to maintain control over experimental conditions,

especially since our participants are geographically dispersed. In addition, we integrated

vignettes into a survey, as this is a promising but rarely applied approach to study

respondents’ judgments and account for the shortcomings of each approach (Atzmüller

& Steiner, 2010). By randomly assigning participants to treatment conditions, we

eliminated systematic differences in the participants that might affect their responses

(e.g., Bachrach & Bendoly, 2011). As a result, differences in response behavior can be

attributed to the manipulated experimental treatments.

3.3.1 Development of vignettes and experimental design

We carefully constructed vignettes to assign the participants to a scenario in which they

served as professional buyers who are considering buying a specific product from a

potential supplier. Thereby, a projective technique – a form of indirect questioning from


the perspective of another person or group – was utilized to limit potential demand

characteristics and effects of social desirability (Fisher, 1993). In line with the principle

of form postponement, all vignettes contained the same introductory paragraph (common

module) to ensure that all participants are provided with a similar contextual background

(Rungtusanatham et al., 2011).

Our factors of interest, CSR reputation and CSIR severity, were manipulated in a

subsequent experimental cues module. All other factors of the vignettes were held

constant. In total, using a 3 (CSR reputation: None, symbolic, or substantive) x 2 (CSIR

severity: Low or high) full factorial design, six vignettes were created. After reading a

vignette, the participants were asked to answer questions regarding their intentions and

perception of the situation. Table 6 shows the descriptions of the vignette modules.

We manipulated our factors of interest as follows. Information on the CSR

reputation of the potential supplier was either not given (none), contained several

symbolic activities (symbolic), or mentioned specific substantive activities (substantive).

To differentiate between substantive and symbolic activities, we relied on Hawn and

Ioannou (2012) and their categorization of 120 Thomson Reuters (ASSET4) items.

Accordingly, a CSR reputation driven mainly by substantive activities was described by

specific policies and quantitative indicators of CSR implementation, whereas claims and

reports represented a CSR reputation driven mainly by symbolic activities.

Each CSR reputation manipulation contained three specific actions. To capture

CSR in a broad sense rather than focusing on one single dimension, each of these actions

addressed one of the three CSR dimensions (environment, society, and governance).

Additionally, both substantive and symbolic CSR reputations were highlighted by

describing that the potential supplier achieved high CSR ratings and that the firms

reported on how they either contribute to the general welfare of society (symbolic) or

implement specific measures to increase the latter (substantive).

In line with previous research, CSIR severity was manipulated by varying the effect

of and damage caused by a specific event (e.g., Germann et al., 2014; Hartmann &

Moeller, 2014). We illustrated the effect or damage associated with a CSIR incident by

altering the number of people that were hurt or affected by a leak of ammonia at one of

the supplier’s facilities. Ammonia leakages frequently occur in practice and vary in terms

of the damage caused (e.g., Wong, 2013). We varied CSIR descriptions to distinguish

between low and high severity.


Table 6: Vignette module text descriptions

Introduction

Mr. Müller is a professional buyer for the textile manufacturer TextileCorp. One of

his responsibilities concerns buying textile printers which allow for a cost-effective

application of color on textiles in specific patterns and designs. Due to difficulties

with the former supplier for textile printers, Mr. Müller has been instructed to find

and select a new one. He compares and analyzes the product offerings of several

potential suppliers and receives corresponding quotations. Based on his analyses, the

firm PrintInc is his favorite choice.

Factor Manipulated factor levels

CSR reputation

Substantive Symbolic None

PrintInc is well-known for its

activities related to corporate social

responsibility (CSR) and receives

very good CSR ratings. To

contribute to the welfare of society,

PrintInc has implemented concrete

and extensive measures. The firm

uses ecological criteria in its

supplier selection process, has an

above-average share of female

supervisory board members, and

pursues a clear strategy to improve

the work-life-balance of its

employees.

PrintInc is well-known for its

activities related to corporate social

responsibility (CSR) and receives

very good CSR ratings. PrintInc

reports about its engagement for the

welfare of society. The firm

publishes a CSR report on an annual

basis, is a member of an initiative to

reduce CO2 emissions, and claims

to strictly control for compliance

with human rights inside of its own

supply chain.

(–)

CSIR severity

High Low

After Mr. Müller identified PrintInc as

his favorite supplier, it became publicly

known that there was an incident at one

of PrintInc’s factories. Due to

insufficient safety precautions, large

amounts of ammonia leaked into the air.

Six employees died at the scene. 25 other

employees, as well as a number of local

residents which lived nearby, had to be

taken to hospital because of massive

respiratory problems and serious

cauterization of their airways.

After Mr. Müller identified PrintInc as

his favorite supplier, it became publicly

known that there was an incident at one

of PrintInc’s factories. Due to

insufficient safety precautions, very

small amounts of ammonia leaked into

the air. Two employees complained of

minor respiratory problems. At no point

of time were residents of this area in

danger.

Note. The table shows translations; the original language was German.

As Wason et al. (2002) recommended, the vignettes were pre-tested with 61

students to assess their validity, internal consistency, and realism. The development of

two different vignette versions ensured that the distinction between substantive and

symbolic CSR was independent of specific semantics or CSR activities and that the CSIR

incident selected was realistic. The two vignette versions differed regarding the depicted

CSIR incident (version 1: Ammonia leak, version 2: Immoral surveillance of employees)

and the delineated CSR activities for substantive and symbolic CSR reputations. The

students were randomly assigned to two vignettes (one out of six per vignette version).


After reading a vignette, the students were supposed to indicate their purchase intention

(PI) and intention to engage in nWOM.

In addition, the students responded to manipulation check items to ensure that the

representations of different levels of CSR reputation and CSIR severity were appropriate.

For both vignette versions, the mean responses differed significantly across different

levels of our manipulated variables. In the pre-test, the students also rated the degree of

realism of the vignettes on a seven-point scale (1 = “not at all” to 7 = “totally”). The

results suggested that the scenarios appeared realistic and the six different vignettes of

version 1 were perceived as slightly more realistic compared to the vignettes of version

2. The mean responses to the six different vignettes of version 1 ranged from 4.8 to 6.1

(5.6 on average) while the means for those of version 2 ranged from 4.4 to 6.0 (5.2 on

average). Consequently, vignette version 1 was selected for data collection. Minor

modifications after the pre-test refined the clarity and wording of manipulations.


Between March and June 2017, the data were collected by means of a self-administered

online experiment. Contact addresses were obtained from a commercial business data

provider. The participants of our experiment were full-time working professionals with

direct experience in supply chain management working for firms from 15 different

industries in Germany, Austria, and Switzerland. The subjects received an invitation via

e-mail and they were randomly assigned to one of the six treatment conditions. Out of the

1064 managers invited, 153 managers (16.3% female) completed the experiment,

resulting in an effective response rate of 14.4%. Inconsistent participation duration led to

the exclusion of five observations. Box plots of the outcome variables showed 13 visible

outliers (Tukey, 1977). In line with prior research (Aguinis, Gottfredson, & Joo, 2013;

Raaijmakers, Vermeulen, Meeus, & Zietsma, 2015), these were excluded from further

analyses resulting in a full sample of 135 usable scenarios. On average, the participants

had almost 17 years of experience in supply chain management (SD = 10.62).

3.3.3 Measures

We focused on two distinct customer outcomes as dependent variables, purchase

intention and intention to engage in nWOM. The measures described in the following


subsection utilize seven-point rating scales (ranging from 1 := “not at all” to 7 :=

“totally”).

Purchase intention (PI). Purchase decisions are typically binary, as one can either

buy or not buy. The use of a binary variable in an experiment, however, results in a

dependent variable with little variance to explain. This would considerably limit the

ability to understand the effects of our independent variables on the participants’

intentions (McKelvie, Haynie, & Gustavsson, 2011). In line with previous research, we

therefore operationalized PI as one of our dependent variables measured on a Likert-type

scale and asked the participants to indicate how likely they would be to buy the respective

product if they were the decision maker described in the scenario (e.g., Sen &

Bhattacharya, 2001; Wilcox, Kim, & Sen, 2009). We thereby followed the

recommendations to use single-item measures for doubly concrete constructs, such as PI

(Bergkvist & Rossiter, 2007). Doubly concrete constructs have both a clear object (e.g.,

a product) and a single-meaning attribute (e.g., willingness to buy).

Intention to engage in nWOM. As a further dependent variable, we measured the

extent to which participants would be willing to engage in nWOM (M = 1.96, SD = 1.23;

α = 0.82) after reading a given scenario (Richins, 1983; Singh, 1990). To this end, the

participants responded to three items, which we adapted to our vignettes from recent

research in the CSR context (Antonetti & Maklan, 2016). The three items reflect an

individual’s willingness to spread negative information about the specific supplier

involved in CSIR.

Given the sensitivity of the investigated topic, and to be able to investigate the

influence of personal attitudes towards CSR on the behavior of the participants, we also

measured their level of CSR support (M = 5.15, SD = 1.12; α = 0.75) using an adapted

five-item scale at the end of the experiment (e.g., Ramasamy, Yeung, & Au, 2010). The

participants had to indicate the degree to which they were (1) willing to pay more to buy

products from a socially responsible firm, (2) considering the ethical reputation of

business when they shop, (3) avoiding to buy products from firms that have engaged in

immoral actions, (4) willing to pay more to buy products of a company that shows

engagement for the well-being of our society, and (5) willing to rather buy from a firm

with a socially responsible reputation, if the price and quality of two products are similar.

Finally, the respondents provided standard demographic information (gender), firm


related information (industry), and their experience in the field of supply chain

management (in years).

The psychometric properties of the two reflective scales (nWOM and CSR support)

were assessed using a covariance-based confirmatory factor analysis (CFA). The

measurement model resulted in an acceptable fit to the data, given the relatively small

sample size (Hair et al., 2009): χ2(19) = 45.76 with p = 0.001 (χ2/df = 2.41), CFI= 0.94,

TLI = 0.91, SRMR = 0.083, and RMSEA = 0.102 (90% confidence interval CI = [0.065,

0.140]). Moreover, all items showed large and statistically significant factor loadings on

their hypothesized factor (p < 0.05 for all loadings). Composite reliability for both

constructs (0.86 for nWOM and 0.75 for CSR support) exceeded the cut-off value of 0.70.

The average variance extracted (AVE) values were 0.67 for nWOM and 0.42 for CSR

support; thus, above or slightly below the 0.5-threshold. Since both constructs extracted

more variance than they share with each other (Pearson correlation coefficient r = –0.10;

r2 = 0.01), discriminant validity was supported. Given these results, the following

analyses use scale averages.

3.3.4 Manipulation checks

As in the pre-test, we verified that our manipulations of the independent variables worked

as intended. To this end, all participants responded to several manipulation check

questions after reading a vignette. The item for CSIR severity asked participants to rate

whether the incident was very severe or not on a seven-point Likert-type scale. The

respondents’ perception of CSIR severity was significantly lower for vignettes that

described a CSIR incident with low severity than for vignettes containing a description

of a CSIR incident of high severity (Mlow = 3.92, Mhigh = 6.58; t(133), p < 0.001). Another

item asked the participants whether they had received information on the supplier’s CSR

reputation before the CSIR incident happened. Almost 84% of the participants who had

received information on CSR correctly responded to this question. If a respondent

answered this question with yes, another item appeared on the screen, which specifically

asked for the perceived nature of the CSR reputation on a seven-point scale (1 := “mainly

symbolic” to 7 := “mainly substantive”). To ensure that all participants had the same

information on how to distinguish substantive from symbolic CSR activities, we provided

a brief explanation to delineate that substantive CSR does involve concrete change of

business practices while symbolic CSR does not. The participants’ responses were


significantly different for substantive and symbolic CSR reputations (Msymbolic = 3.97,

Msubstantive = 4.77; t(65), p = 0.04).

3.4 Results

To test our hypotheses, we relied on analysis of variance (ANOVA). Parametric tests

(e.g., ANOVA) are appropriate and robust for rating-scale data, and they can be used with

“small sample sizes, with unequal variances, and with non-normal distributions, with no

fear of ‘coming to the wrong conclusion’” (Norman, 2010, p. 631). We conducted two

two-way ANOVAs with CSR reputation (none, symbolic, or substantive) and CSIR

severity (low or high) as between-subjects factors. Table 7 shows the number of

participants per cell, the cell means, the corresponding standard deviations for both of our

dependent variables PI and nWOM, and the correlation between our dependent variables.

Table 8 and Table 9 show the results of the two ANOVAs.

Table 7: Frequencies, cell means, standard distributions, and correlation for the

dependent variables

Experimental condition Purchase

intention (PI)

Negative word-of-

mouth (nWOM) Number of

observations

(n = 135)

Correlation of PI

and nWOM

(roverall = –0.48) CSIR

severity

CSR

reputation M SD M SD

Low

None 4.52 1.97 2.06 0.96 23 –0.28

Symbolic 6.47 0.62 1.10 0.16 17 –0.50

Substantive 5.73 0.94 1.33 0.43 22 –0.16

High

None 3.21 2.09 2.21 1.25 33 –0.35

Symbolic 3.33 2.13 2.84 1.90 21 –0.47

Substantive 3.58 2.14 1.96 1.00 19 –0.27

Table 8: ANOVA results with purchase intention (PI) as dependent variable

Source Partial SS df MS F p-value

CSR reputation 27.49 2 13.74 4.19 0.017 *

CSI severity 156.29 1 156.29 47.65 0.000 ***

CSR reputation × CSI severity 18.59 2 9.30 2.83 0.062 †

Model 194.51 5 38.90 0.62 0.000 ***

Residual 423.15 129 3.28

Note. SS refers to “sum of squares”, df to “degrees of freedom”, and MS to “mean square”; n = 135.

† p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

PI was the dependent variable in the first ANOVA. The results revealed an only

marginally statistically significant interaction effect between CSR reputation and CSIR

severity (F = 2.83, p = 0.06). The main effects of CSR reputation (F = 4.19, p = 0.02)


and CSIR severity (F = 47.65, p < 0.001) were statistically significantly different from

zero, but are qualified by the interaction effect; hence, we do not dwell on them.

Table 9: ANOVA results with intention to engage in negative word-of-mouth (nWOM)

as dependent variable


CSR reputation 5.55 2 2.77 2.18 0.118

CSI severity 22.98 1 22.98 18.03 0.000 ***

CSR reputation × CSI severity 14.22 2 7.11 5.58 0.005 **

Model 39.89 5 7.98 6.26 0.000 ***

Residual 164.39 129 1.27

Note. SS refers to “sum of squares”, df to “degrees of freedom”, and MS to “mean square”; n = 135.

† p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

Planned contrast analysis revealed that when CSIR severity was low, PI was higher

for substantive and symbolic CSR reputations than for no CSR reputation (Msubstantive, low

= 5.73, Msymbolic, low = 6.47, Mnone, low = 4.52; p’s < 0.05), supporting Hypotheses 1a and

2a. Furthermore, when CSIR severity was high, the effect of PI was not significantly

different from zero across different CSR reputation types (Msubstantive, high = 3.58,

Msymbolic, high = 3.33, Mnone, high = 3.21; p’s > 0.05), providing no support for Hypothesis 3a.

Hypothesis 4a, which proposed that symbolic CSR aggravates negative reactions to

severe CSIR, was also not supported.

The second ANOVA used nWOM as the dependent variable. Similarly, the results

revealed a strong and statistically significant interaction between CSR reputation and

CSIR severity (F = 5.58, p = 0.005). The main effect of CSIR severity (F = 18.03, p <

0.001), which was qualified by the mentioned interaction effect, was also statistically

significantly different from zero. Planned contrasts examined differences between

specific groups. In line with Hypotheses 1b and 2b, when CSIR severity was low, nWOM

was significantly lower for substantive and symbolic CSR than for no CSR (Msubstantive, low

= 1.33, Msymbolic, low = 1.10, Mnone, low = 2.06; p’s < 0.05). When CSIR severity was high,

nWOM was not significantly different between the substantive and no CSR conditions

(Msubstantive, high = 1.96, Mnone, high = 2.21; p = 0.45). This is inconsistent with Hypothesis

3b. Finally, as predicted in Hypothesis 4b, nWOM was significantly higher for symbolic

CSR than for no CSR (Msymbolic, high = 2.84, Mnone, high = 2.21; p = 0.05). Figures 6a and 6b

depict the two interaction effects, and Table 10 provides a summary of our tested

hypotheses.


Figure 6: Interaction effect of CSR reputation and CSIR severity

(a) Purchase intention (PI) as dependent

variable

(b) Intention to engage in negative word-

of-mouth (nWOM) as dependent

variable

Table 10: Summary of hypotheses and results

Hypothesis Prediction Result

H1

If a firm with a reputation for substantive CSR is involved in non-severe

CSIR, it experiences smaller negative effects on its customers’

(a) purchase intention Supported

(b) intention to engage in negative word-of-mouth Supported


H2

If a firm with a reputation for symbolic CSR is involved in non-severe CSIR,

it experiences smaller negative effects on its customers’

(a) purchase intention Supported



H3

If a firm with a reputation for substantive CSR is involved in severe CSIR,

it experiences smaller negative effects on its customers’

(a) purchase intention Not supported

(b) intention to engage in negative word-of-mouth Not supported


H4

If a firm with a reputation for symbolic CSR is involved in severe CSIR,

it experiences larger negative effects on its customers’

(a) purchase intention Not supported



To examine the robustness of our findings and to ensure that the participants’

intentions were not determined by their level of CSR support and experience, we included

both variables as covariates in separate analyses of covariance (ANCOVA) with PI and

nWOM as dependent variables and CSR reputation as well as CSIR severity as

independent variables. Neither the level of CSR support nor the work experience in

4.52

3.20

6.47

3.33

5.73

3.58

1

2

3

4

5

6

7

low high

Pu

rcha

se

inte

ntio

n(P

I)

CSIR severity

no CSR

symbolic CSR

substantive CSR

2.06 2.21

1.10

2.84

1.33

1.96

1

2

3

4

5

6

7

low high

Inte

ntio

n to

en

gage

in n

egative

wo

rd-o

f-m

ou

th(n

WO

M)

CSIR severity

no CSR

symbolic CSR

substantive CSR


supply chain management of the participants had a statistically significant influence on

the dependent variables. Hence, the predicted effects remained qualitatively similar.

3.5 Discussion

In the academic literature, the topic of CSIR has rarely been addressed, although

“avoiding bad” constitutes a precondition for firms to be perceived as socially

responsible. The recurrent examples of firms that are involved in CSIR prove that it is a

challenging task to prevent environmental and social misconduct. Subsequent negative

stakeholder reactions may severely damage a firm’s performance and reputation; thus,

responsible managers need to address them. Many firms believe that their ex ante CSR

engagement may provide insurance-like effects, which attenuate negative effects of

CSIR. Several research efforts have supported this idea. However, we demonstrated that

a firm’s reputation for CSR plays a more complex role in managing stakeholder reactions

to CSIR. Thereby, we contribute to the literature on CSR, CSIR, and disruption

management by (1) providing insights into the moderating effects of ex ante CSR

reputations in case of CSIR, (2) demonstrating that the insurance-like or aggravating

effects of prior CSR reputations are determined by the reputation’s nature (substantive/

symbolic) and the severity of CSIR, and (3) improving our understanding of managing

CSR and CSIR in buyer-supplier relationships.

3.5.1 Theoretical implications

This study analyzed the role of a firm’s CSR reputation in moderating negative

stakeholder reactions to CSIR. Our analysis, based on the data from a vignette-based

experiment with supply chain professionals, yielded several important theoretical

implications. We addressed a number of recent calls for more research on the topic of

CSIR (e.g., Lin-Hi & Blumberg, 2018; Lin-Hi & Müller, 2013), CSR-related crisis

management (e.g., Janssen et al., 2015; Lenz et al., 2017; Shiu & Yang, 2017), the

implications of utilizing substantive and symbolic CSR actions (Hawn & Ioannou, 2012,

2016), and industrial buying contexts (Homburg, Stierl, & Bornemann, 2013) to fill an

important and relevant gap in the current academic discussion on the value of ex ante

CSR reputations at times of crises.

First, we contribute to the body of literature that explicitly distinguishes between

CSR and CSIR as two different constructs. This distinction is central to our research


design, as CSR comprises voluntary activities of a firm to “do good” while CSIR reflects

a firm’s inability to “avoid bad.” Our findings address the relationship between CSR and

CSIR and provide insights into their effect on the intentions of stakeholders. More

specifically, as suggested by prior research (e.g., Lin-Hi & Blumberg, 2018; Lin-Hi &

Müller, 2013), our results highlight that preventing (severe) CSIR is a precondition for

firms to effectively utilize their CSR reputation.

Second, and in line with previous research, our study highlights that it is important

to distinguish between different facets of CSR (Hawn & Ioannou, 2012, 2016). Moreover,

this study’s results provide empirical evidence that the distinction between substantive

and symbolic CSR is relevant not only in business-to-consumer (B2C) but also in B2B

contexts. The circumstance that our result do not provide support for Hypotheses 3a and

3b might have been driven by compliance concerns and issues from the participants’ own

work environments. Such concerns might have affected the results in a way that the more

favorable attribution of substantive CSR were not sufficient to mitigate the perceived

negativity of the presented CSIR incident. This would highlight an important difference

between consumers and professional buyers.

Third, beyond adding further empirical support to the observation that insurance-

like or risk-mitigating effects of CSR do not always hold (e.g., Sen & Bhattacharya,

2001), we contribute to specifying the boundary conditions of these effects and

demonstrate that whether a firm benefits from “doing good” or not (when it is involved

in CSIR) depends on the nature of a firm’s CSR reputation (substantive/ symbolic) and

CSIR severity. The results suggest that both substantive and symbolic CSR reputations

mitigate negative stakeholder reactions to corporate environmental and social misconduct

if the severity of a specific CSIR incident is low. When firms are involved in severe CSIR

the effect of ex ante CSR can switch from insulation to amplification. Prior research has

found that severe CSIR incidents can recalibrate stakeholders’ expectations (Huq,

Chowdhury, & Klassen, 2016). Our results add to this insight by demonstrating that in

case of severe CSIR, the effect of a firm’s CSR reputation on negative stakeholder

reactions depends on its nature and the specific dimension of the stakeholder reaction.

While firms with substantive CSR reputations seem not to benefit from their ex ante

activities regarding negative stakeholder reactions to severe CSIR, stakeholders of firms

with reputations driven mainly by symbolic CSR actions even express higher intentions

to engage in nWOM compared to stakeholders of firms that have no reputation for CSR.


For PI, we were not able to find sufficient statistical evidence to show that the

reactions of stakeholders to severe CSIR depend on the CSR reputation. This must not be

interpreted as evidence that there is no effect. Methodological reasons might have

influenced this result. For example, in case of symbolic CSR, contrast effects resulting in

a change of both PI and nWOM might require higher levels of CSIR severity. Similarly,

the chosen description of high severity might already represent a level too high for

substantive CSR to result in assimilation effects. The incident described in our

experimental material might have been too severe to be attenuated by favorable

attributions of a firm’s CSR motives. However, a possible explanation for the lack of

evidence to support Hypothesis 4a is that for an individual’s PI, it might not be relevant

whether substantive or symbolic CSR engagement has formed stakeholder expectations.

In case of severe CSIR, the disparity between a customer’s expectations and a firm’s true

performance might be perceived as similar for both substantive and symbolic actions

since they can also be equally successful in gaining legitimacy and fostering expectations.

Customers might be more willing to share the disconfirmation of their expectations if

firms used symbolic actions instead of no CSR engagement to affect their expectations

based on the perceived motives of a firm’s CSR engagement. Once ex ante CSR efforts

are identified as greenwashing, they cast doubt on a customer’s prior expectations. To

prevent others within their peer group from building high expectations based on symbolic

CSR engagement, customers seem to be more willing to engage in nWOM than if the

expectations were based on substantive CSR or no CSR engagement at all.

Finally, our findings enhance our understanding of the impact of CSR on industrial

buying situations. To the best of our knowledge, this is the first study that explicitly

addresses the role played by ex ante CSR reputations on CSIR-related disruption

management in a B2B environment. Previous research has highlighted that many decision

makers might not be aware of the benefits associated with CSR-related activities along

their firms’ supply chains (Paulraj, Chen, & Blome, 2017), but we provide empirical

support for implementing CSR based on instrumental rather than moral motives.

Moreover, prior studies have focused on the B2C perspective and important consumer

outcomes, such as satisfaction and loyalty. Although these findings may partially hold

true in B2B contexts, it is well-researched that professional buyers differ considerably

from consumers. Thus, we focused on customer outcomes, considered characteristics of

industrial buying, and addressed a highly relevant topic because our research is among


the first studies to empirically analyze positive effects of CSR actions in industrial buying

contexts (Homburg et al., 2013). As firms are increasingly held responsible for

misconduct within their supply chain, the implications of our study extend previous work

by emphasizing that not only firms which are positioned downstream the supply chain

can considerably benefit from CSR but also those which are further upstream and less

visible to consumers (C. G. Schmidt, Foerstl, & Schaltenbrand, 2017).

3.5.2 Managerial implications

Several implications for managerial practice can be deduced from the results. Given the

study’s focus on industrial buying contexts, a key insight is that CSR engagement is

worthwhile not only in B2C but also in B2B contexts. Both substantive and symbolic

CSR reputations can have insurance-like effects in case of non-severe CSIR in industrial

buying situations. This provides a strong justification for managers to pursue an

intensified engagement in CSR to benefit from mitigated negative stakeholder reactions

in case of misconduct. In addition, the study’s findings underline the necessity of

managers to proactively reflect on different stakeholders and their expectations, as

delineated in previous work (Gualandris, Klassen, Vachon, & Kalchschmidt, 2015).

However, our results suggest that engaging in CSR without considering the nature of the

specific actions can backfire in times of crises. Customers can distinguish between

substantive and symbolic actions, which has important repercussions on their behavior.

CSR reputations based on symbolic actions cannot fully avoid more intense negative

stakeholder reactions when a firm is involved in severe CSIR. Thus, managers should be

aware of the nature of their firm’s own CSR engagement. Prior research highlights that

the decision to pursue substantive or symbolic CSR activities depends on cost-benefit

considerations (e.g., Christmann & Taylor, 2006; Durand et al., 2017; Kaul & Luo, 2018).

However, based on our results, if the expected probability of severe environmental or

social misconduct is considerably high, managers should be careful about addressing

stakeholder demands with potentially backfiring symbolic CSR actions. In this case,

managers might rather decide to invest in substantive CSR activities instead. Compared

to merely symbolic CSR claims that do not entail specific changes in business processes,

substantive CSR has the potential to mitigate environmental and social risk while offering

insurance-like effects if a firm is involved in non-severe CSIR.


In addition, our findings suggest that managers who decide to engage merely in

“doing good” to develop a firm’s reputation for CSR might not always benefit from a

competitive advantage over firms that provide similar products but do not have a

reputation for CSR. When (potential) customers are confronted with a supplier’s

involvement in severe CSIR, firms with CSR reputations driven by substantive CSR

actions seem to perform just as good as firms that do not have a reputation for CSR. Firms

that engage mainly in symbolic actions might even have a disadvantage and suffer more

than do firms that do not promote themselves as socially responsible. This underlines the

need for managers to also consider “avoiding bad” as a precondition to benefit from

voluntary CSR engagement. Hence, more attention, effort, and resources should be

invested in mitigating the risk that CSIR occurs within their firms’ supply network to

effectively utilize a firm’s reputation for CSR, since engaging in CSR does not

automatically provide protection against negative stakeholder reactions.


This study and its findings are subject to certain limitations, which provide fruitful

avenues for future research. The participants examined scenarios containing information

about a specific one-shot buying situation describing a potential supplier’s CSR

reputation and involvement in CSIR. In real industrial buying processes, the participants

are likely to receive such information in a more fragmented fashion, perhaps over multiple

time periods. Moreover, we focused on individuals with centralized decision making

authority and one single supplier option to choose from, which might often deviate from

organizational practice. Thus, further research should account for the dynamics of

organizational buying processes and buying decisions of teams while considering

multiple supplier options would add to further support the validity and generalizability of

our findings. In addition, field experiments investigating the effect of CSR reputations on

stakeholder reactions to CSIR would also contribute an increased generalizability of our

findings, although we carefully designed our experiment to provide externally valid

results. This is especially important, because the use of vignettes enabled us to analyze

only the participants’ intentions rather than their actual behaviors. Still, ample empirical

evidence suggests that intentions may serve as reliable indicators of actual behavior (e.g.,

Ajzen, 1991; T. L. Webb & Sheeran, 2006).


In the manipulations of CSR reputations, we have focused on describing firms with

and without a reputation for CSR. This means that we kept the “strength” of a reputation

for CSR constant. Future research might investigate whether the effects described in our

study can be replicated with reputations of different strength. To generate insurance-like

effects, reputations for CSR might have to exceed a certain threshold of power. Prior

research, for instance, has highlighted that short histories of CSR engagement are not

sufficient to mitigate negative stakeholder reactions to CSIR (Vanhamme & Grobben,

2009).

This study took a single episode perspective, although firms might be involved in

several recurring incidents of environmental and social misconduct over time. A

promising direction for future research is to investigate whether and how a reputation for

CSR immunizes against CSIR in multi-period settings. For example, initial insurance-

like effects of reputations for CSR could end up having an aggravation effect in case of

recurring involvement in CSIR. Customers might tolerate minor isolated performance

deviations from their expectations, but once a firm is involved in several cascading CSIR

incidents, its CSR reputation might not provide protection anymore and could, on the

contrary, lead to aggravated negative reactions. Just recently, Shiu and Yang (2017)

reported diminishing insurance-like effects of CSR. However, as they did not distinguish

between substantive and symbolic CSR engagement and focused on investors, some open

questions regarding the evanescence of the insurance-like mechanism of ex ante CSR

remain.

Moreover, the effects of substantive and symbolic CSR on negative stakeholder

reactions might be closely related to trust. Trust might affect an individual’s effort used

to distinguish substantive from merely symbolic CSR while a customer’s perception of a

firm’s CSR might influence the effect of service failure on customer trust (Bozic, 2017;

Choi & La, 2013). Hence, exploring how trust affects the moderating role of CSR

regarding the link between CSIR and negative stakeholder reactions is a promising

avenue for future research.

Another interesting avenue for future research could be to investigate whether

professional buyers differ considerably from consumers regarding how they react to firms

with a reputation for CSR that are involved into CSIR. On the one hand, consumers might

react even more negatively to severe misconduct, regardless of the nature of a firm’s CSR

reputation, especially if switching costs are low. On the other hand, whether a large


disparity between expectations and a firm’s true performance will be tolerated might

depend on the loyalty of a consumer. In addition, the behavior of professional buyers is

subject to corporate governance rules and standard operating procedures that determine

compliant behavior. These constraints might limit the response options or even dictate a

specific behavior. This aspect has not been considered in our experimental design.

Another limitation of this study is the relatively moderate sample size. Of special

interest for future research could be the potential boomerang effects regarding Hypothesis

4b. This study indicates that symbolic CSR engagement can potentially backfire

following CSIR but it does not analyze the underlying cause of this effect in more detail.

More specifically, subsequent studies might investigate why symbolic CSR merely led to

more negative stakeholder reactions concerning nWOM but not PI. In addition and

closely related to the aforementioned issue, for PI as dependent variable, conclusions

based on our results need to be drawn with caution due to the limited evidence provided

by the only marginally significant interaction effect of CSR reputation and CSIR severity.

Hence, to achieve a higher generalizability regarding the generated insights, we

encourage the validation of our results using a larger sample.

Finally, choice behavior might also vary with the cultural background of

respondents. We chose to limit the empirical study to Germany, Austria, and Switzerland,

because from a cultural perspective, these three countries are usually considered to be

very similar and homogeneous (Hofstede, 1984, 2003). A relevant next step would be to

scrutinize the robustness of the effects across different cultural regions. People from

North America, for example, differ from the population sampled in this study in that they

display lower uncertainty avoidance and higher degree of individualism. It could be

conjectured that North American supply chain professionals are less affected by the

uncertainty created by the discrepancy between a firm’s CSR reputation a case of

misconduct (lower uncertainty avoidance), although at the same time, they could be less

willing to engage in negative word-of-mouth (higher degree of individualism).

Supply disruption management: The early bird catches the worm, but the second mouse gets the cheese? 61

Chapter 4 Supply disruption management: The

early bird catches the worm, but the

second mouse gets the cheese?

Co-author:

Christoph Bode


Abstract6

In the direct aftermath of supply disruptions, managers typically face uncertainty in the

form of incomplete or unreliable information about the consequences of possible response

actions. To gain a better understanding of how to recover from disruptions and support

effective disruption management, this study pursues a multi-method research approach.

First, we propose an agent-based simulation model of supply disruption responses and

recovery processes to reveal how managers should behave. Based on this model, we

analyze the performance outcomes of a decision maker’s tendency to reduce uncertainty

by collecting further information before taking action (ready-aim-fire) or to act

immediately (ready-fire-aim) under different conditions. Second, vignette-based

experiments with supply chain professionals are conducted to show how managers

actually intend to behave in supply disruption recovery. The findings suggest that

managers' intentions deviate from how they should behave and that quick reactions to

disruptions can be beneficial, even if the exact consequences of a response action cannot

precisely be determined. Furthermore, in complex environments, ready-fire-aim leads to

a better average performance than ready-aim-fire, if response uncertainty is not very high.

The derived insights provide novel theoretical and managerial implications for effective

disruption management.

6 Merath, M. and Bode, C., 2018. Supply disruption management: The early bird catches the worm, but the

second mouse gets the cheese? Unpublished Working Paper, 1-45. An earlier version won the “Chan Hahn

Best Paper Award” of the Operations and Supply Chain Management Division of the Academy of

Management at the Annual Meeting in Atlanta, GA, in 2017.


4.1 Introduction

Supply chain disruptions are defined as “unplanned and unanticipated events that disrupt

the normal flow of goods and materials within a supply chain […] and, as a consequence,

expose firms within the supply chain to operational and financial risks” (Craighead et al.,

2007, p. 132). In the last decades, firms’ risk of being harmed by disruptions seems to

have increased (Sodhi, Son, & Tang, 2012). On the one hand, modern supply chains have

evolved into more complex and vulnerable interconnected networks (Bode & Wagner,

2015). On the other hand, natural disasters are becoming more frequent (Munich Re,

2015).

Many types of supply chain disruptions – especially those characterized by high

impact and low probability – cannot be completely avoided. The decisions that managers

take in response to severe disruptions inevitably affect the success of a firm’s recovery,

as demonstrated by the often-cited Albuquerque fire (Latour, 2001). On March 17, 2000,

a Philips plant in Albuquerque, New Mexico, was hit by lightning and caught fire.

Initially, it seemed that the damage would be limited. Philips informed the two main

customers of the chips produced in this plant, Nokia and Ericsson, about an expected

delivery delay of one week. Yet, the responses of the two competitors were completely

different. Nokia reacted directly, put pressure on Philips, and collaborated with

alternative suppliers to rebound from the disruption with minimal negative consequences.

Ericsson adopted a “wait and see” approach and delayed remedial action until more

information became available. When it was clear that the damage to the clean rooms was

far more substantial than expected, Ericsson was not able to find an alternative supplier

on short notice and had to delay the launch of a new product while losing market share to

Nokia (Latour, 2001; Sheffi, 2005).

The possible losses that supply chain disruptions may entail are well-researched

(Hendricks & Singhal, 2005a,b), but the process of reactive supply disruption

management has received only limited research attention. In particular, our understanding

of how firms should respond to supply chain disruptions in order to quickly recover is

weak (Bode et al., 2011; Macdonald & Corsi, 2013). This issue is important, because a

firm’s ability to effectively respond to sudden disruptions is critical to both its short-term

performance and long-term competitiveness.


Responding to disruptions usually requires decision making in complex

environments characterized by uncertainty and limited information. Under these

conditions, some firms defer actions until (most of) the uncertainty has been resolved and

reliable information is available, while other firms respond immediately, even though the

information at hand is cloudy (Kleinmuntz & Thomas, 1987). Using the analogy of

shooting, the first approach can be termed ready-aim-fire (RAF) and the second ready-

fire-aim (RFA). Intuitively, RAF may lead to more precise actions than RFA, but loss is

typically a function of time and RAF may also allow quicker competitors to obtain

superior positions. Conversely, RFA carries the inefficiencies of trial and error and, when

decisions are irreversible and path-dependent, the risk of pursuing an inferior recovery

path. Hence, the key question is: Under which circumstances should a manager delay its

response actions until more evidence is acquired and how do managers actually intend to

behave?

This study addresses this issue by means of a multi-method research approach.

First, we develop an agent-based model to perform simulation experiments and delineate

how managers should theoretically behave in disruption response situations. Second, we

conduct vignette-based experiments with supply chain managers to adopt a more

behavioral perspective and demonstrate how managers actually intend to approach

disruption recovery as opposed to how they should behave. The results provide insights

into effective decision making in the aftermath of supply chain disruptions and contribute

to the knowledge of supply chain disruptions management.

4.2 Supply disruption management and resilience

The complexity and interconnectedness of modern supply chains create vulnerabilities

that expose firms to the risk of being affected by severe disruptions of the flows of

materials, information, and funds (Sheffi, 2005). A supply chain disruption is a

combination of an unforeseen event that interrupts these flows and a subsequent situation

that distorts the normal course of a focal firm’s business operations. We focus on

disruptive events that cannot be resolved in the course of daily operations management

and that occur in a focal firm’s upstream supply chain (in the following: Supply

disruptions). It has been suggested that supply chain disruptions “are more critical when

they occur upstream in the chain” (Pereira et al., 2014, p. 627).


There is a large body of literature on these disruptions and the management of

supply chain risks (e.g., Heckmann et al., 2015; Rao & Goldsby, 2009). In these works,

there is an agreement that supply disruptions have a certain time profile regarding their

adverse effects on firm performance (Sheffi, 2005). As shown in Figure 7, a disruption

leads to a sudden drop in operating performance which then triggers search and recovery

efforts (Bode et al., 2011).

Figure 7: Typical supply disruption profile

Although supply disruptions may have serious negative consequences for firm

performance, many firms prove themselves unprepared (Handfield, Blackhurst, Elkins,

& Craighead, 2007). The ability to recover from disruptive events is a major component

of a firm’s resilience, a concept which has been widely used in numerous disciplines.

Resilience is the “ability of a system to return to its original state or move to a new, more

desirable state after being disturbed” (Christopher & Peck, 2004, p. 2). In the supply chain

context, Sheffi (2005) defined resilience as the ability and the speed at which firms fully

recover (i.e., return to normal performance levels) from high-impact/ low-probability

disruptions. He added that when “thinking about resilience, it may not be productive to

think about the underlying reason for the disruption – the kind of random, accidental, or

malicious act that may cause a disruption. Instead, the focus should be on the damage to

the network and how the network can rebound quickly” (p. 14). For this reason, this

investigation does not focus on the causes of supply disruptions but solely on their impact

on operating performance.


4.3 How managers should behave in supply disruption

response situations

4.3.1 A model of supply disruption recovery

To recover from a disruption, managers need to actively adjust their firms’ operations

(Blackhurst, Craighead, Elkins, & Handfield, 2005; Chakravarthy, 1982), which can be

interpreted as an adaptation process. Our model of this process is based on the NK model

(Kauffman, 1993; Levinthal, 1997) and is similar to recent organizational research using

agent-based modeling (e.g., Chandrasekaran, Linderman, Sting, & Benner, 2016;

Siggelkow & Rivkin, 2005). Many models of organizational behavior focus on closed-

form solutions that maintain analytical tractability by substantially simplifying the

representations of organizations their environments. Agent-based modeling allows us to

consider more complex settings than closed-form approaches and to generate meaningful

hypotheses as a basis for empirical studies (Siggelkow & Rivkin, 2005). Following

Burton and Obel (1984, 1995, 2004), we build a model that is just as complex as

necessary. It “should not be seen as a literal representation of environments and

organizations, but as the simplest representation that can fulfill our intended purpose”

(Siggelkow & Rivkin, 2005, p. 103).

Fundamental to our study is to view firms as information-processing systems

(Galbraith, 1977; Thompson, 1967; Tushman & Nadler, 1978). According to

information-processing theory (IPT), responses of organizations to environmental

changes are driven by sequential information-processing activities (Barr, 1998; Dutton,

Fahey, & Narayanan, 1983). IPT has been widely used to study organizational decision

making processes subsequent to exogenous changes (e.g., Bode et al., 2011). Most

notably, Mintzberg, Raisinghani, and Theoret (1976) identified general steps of strategic

decision making that form an iterative incremental process: Identification, development,

and selection. Hale, Hale, and Dulek (2006) transferred this process model to crisis

responses and found that recognition (problem discovery and identification of a need to

take a decision), search (for information and alternative actions), and evaluation/ choice

of response options are common crisis recovery stages. Since crisis response processes

are comparable to disruption response processes, we consider these stages part of a firm’s

disruption response process.


4.3.2 The environment

We assume that each decision maker faces a choice problem consisting of N different

binary decisions. These decisions represent the configuration of the firm’s activities and

determine the overall performance given the environment. Examples of such decisions

include the selection of a supplier, the logistics and delivery concept, or whether a specific

material can be substituted or not. A specific instance of a firm’s choices is called choice

configuration and denoted by the decision vector d = (d1, …, di, …, dN) ∊ {0, 1}N. The

choice configurations are evaluated by a fitness value function F = F(d) which expresses

the firm performance the decision makers seek to maximize.

The contribution of each decision i to the fitness value depends not only on its own

state di (0 or 1) but also on the complexity of the environment. A large number of

decisions (N) does not make a problem complex per se, because complexity emerges from

elements of a system “that interact in a nonsimple way” (Simon, 1962, p. 468). We regard

these interdependencies as characteristics of a firm’s environment, because they “are

dictated by the nature of the decisions themselves and […] are not chosen by the firm”

(Siggelkow & Rivkin, 2005, p. 103). The complexity of an environment is expressed by

the intensity of interactions among the N decisions specified by the parameter K ∊ {0, 1,

…, N – 1}. K determines the number of decisions upon which the contribution of each

decision i depends. We denote the state of these K decisions by the vector d–i = (di1, …,

diK) ∊ {0, 1}K. The fitness contribution of each decision i is evaluated by a contribution

function Ci = Ci(di, d–i). If K equals 0, Ci depends solely on the state of di. If K equals N

– 1, Ci depends on di and the states of all other decisions.

After N and K have been specified, a pattern of interaction among the decisions is

created. K decisions are randomly assigned to each decision i. For each of the possible

2K+1 realizations of di and d–i, a contribution Ci is generated by drawing a random number

from a standard uniform distribution (i.e., Ci ~ U(0, 1)). This is repeated for all N

decisions. The fitness value of a choice configuration d is then defined as the average of

the contributions of each decision:

𝐹(𝐝) =∑ 𝐶𝑖(𝑑𝑖; 𝐝−𝑖)𝑁

𝑖=1

𝑁 . (7)

All 2N possible choice configurations and their respective fitness values form a

firm’s environment (e.g., Chandrasekaran et al., 2016; Levinthal, 1997). Interactions


among decisions cause environments to become “rugged” and multipeaked instead of

smooth and equipped with single optima. If K equals 0, a change of a single decision

leaves the contributions of all other decisions unaffected and the fitness value can always

be improved by changing each decision to the state with the highest contribution. If K is

larger than 0, environments become “rugged” in a sense that changing the state of one

decision from 0 to 1 or vice versa also affects the fitness contributions of K other

decisions, leading to multiple local optima. The number of the latter tends to increase

with K, making the search process for a high peak increasingly difficult.

4.3.3 Organizational adaptation to environmental change

Firms adapt to their environment by means of sequential dynamic search processes and

corresponding information processing (e.g., Ethiraj & Levinthal, 2004; Katila & Ahuja,

2002; Levinthal, 1997; March & Simon, 1958). Due to the bounded rationality of decision

makers and firms, the alternatives of how to adapt to environmental changes are not

entirely and instantly known and must first be searched or discovered (Simon, 1955).

These search processes are guided by “satisficing” rather than by optimizing (Simon,

1979).

The behavioral theory of the firm suggests that organizational search is problem-

oriented and triggered by performance shortfalls (Cyert & March, 1963). Furthermore,

firms tend to search locally for alternative actions (March & Simon, 1958; Winter,

Cattani, & Dorsch, 2007). They modify their configuration by identifying and

implementing superior configurations which are part of the immediate “neighborhood of

the organization’s current practices” (Knudsen & Levinthal, 2007, p. 43). In line with

previous research on NK models, we follow the central assumption of local search (e.g.,

Gavetti, Levinthal, & Rivkin, 2005; Winter et al., 2007). During our simulation

experiments, firms examine alternative choice configurations which differ from their

current choice configurations in the state of one of the N decisions, selected at random. If

it provides a greater fitness value than the current choice configuration, the state of the

respective decision is changed (i.e., from 0 to 1 or vice versa). Otherwise, the choice

configuration remains unchanged. This procedure is repeated in every period and often

termed local “hill climbing” (Levinthal, 1997).

An environmental change can have various causes and take many forms. When it

comes to events such as supply disruptions, firms tend to centralize and assign decision


making responsibility to a single manager (“troubleshooter”) or team (Dubrovski, 2004).

Accordingly, agents in our simulation model can be seen as individual decision makers

or teams that have been tasked with the recovery decisions. Their behavior is predefined

by a set of rules.

4.3.4 Supply disruptions and uncertainty

Uncertainty is a central concept for research concerned with the relationship between

organizations and their environments (e.g., Dill, 1958; Duncan, 1972; Thompson, 1967).

In congruence with the information-processing perspective, we build on the notion that

uncertainty and information are closely interrelated. Uncertainty is defined “as a

manifestation of some information deficiency, while information is viewed as the

capacity to reduce uncertainty” (Klir, 2005, p. xiii). For decision makers, uncertainty is

the “difference between the amount of information required to perform a task and the

amount of information already possessed” (Galbraith, 1973, p. 5). Gathering information

to reduce uncertainty is a critical activity for decision making (Cooper, Folta, & Woo,

1995).

Three types of environmental uncertainty can be distinguished: State, effect, and

response uncertainty. State uncertainty involves the inability to understand or predict the

state of the environment, while effect uncertainty impairs an individual’s ability to predict

the impact of environmental changes on organizations (Milliken, 1987). For actions,

response uncertainty is the most prevailing issue (McMullen & Shepherd, 2006). It is

defined as a “lack of knowledge of response options and/or an inability to predict the

likely consequences of a response choice” (Milliken, 1987, p. 137) and is salient in almost

every supply disruption recovery process. It is experienced when there is a perceived need

to act or formulate “a response to an immediate threat in the environment” (Milliken,

1987, p. 138). Especially in the early stages of a disruption response, information on

causes and effects tends to be incomplete or inaccurate (M. Chen, Xia, & Wang, 2010).

From a behavioral perspective, uncertainty is associated with “a sense of doubt that

blocks or delays action” (Lipshitz & Strauss, 1997, p. 150). The potential costs of an

incorrect response may deter a decision maker from taking action (Milliken, 1987).

However, delaying a decision is also costly and may allow competitors to respond faster

to disruptions. Research on disruption management has provided anecdotal evidence that

the disruption recovery performance is time-dependent (Macdonald & Corsi, 2013). Yet,


we still lack convincing evidence on the link between response speed and recovery

performance.

In previous research on organizational search behavior, the problem of evaluating

response alternatives has been treated as being trivial (Knudsen & Levinthal, 2007). To

address this criticism, we investigate the impact of response uncertainty on the evaluation

of response alternatives. During every simulation run, the environment is replaced by a

new randomly generated environment to represent the effect of a supply disruption.

However, the latter is accompanied by response uncertainty and fitness values of

alternative configurations are distorted by a uniformly distributed random error. The

firms differ in the way they react to this response uncertainty as described in the next

sections.

4.3.5 Disruption response strategies

Search processes of firms are affected by preferences and characteristics of individuals.

Decision makers may differ in the degree of risk avoidance (e.g., depending on cultural

differences (Hofstede, 1993)) or the preferred amount of information collected to solve a

problem (O'Reilly, 1982). Accordingly, some organizations penetrate their environment

more than others. Daft and Weick (1984) noted that “organizations may leap before they

look” (p. 288). Against this backdrop, we distinguish two disruption response strategies

under response uncertainty: Ready-aim-fire and ready-fire-aim. These strategies7 have

been mentioned in the fields of management, strategy, and decision theory (e.g., Cox,

2000; Peters & Waterman, 1982), but, to the best of our knowledge, have not been clearly

delineated.

Ready-aim-fire (RAF). In the face of uncertainty, some firms may prefer to delay

a decision until more information becomes available (Kleinmuntz & Thomas, 1987). A

firm can receive additional information passively (e.g., via media announcements) or

actively (e.g., through its own investigations) (Milliken, 1987). Action is delayed in the

hopes that less uncertainty makes the choice of an appropriate alternative easier and more

accurate. Borrowing an analogy from marksmanship, we call this strategy ready-aim-fire.

Its main idea is to reduce response uncertainty and purse accurate analysis before action

is taken (Lipshitz & Strauss, 1997; Mintzberg & Westley, 2001; Peters & Waterman,

7 Similar terms that can be found in the literature are “look before you leap” vs. “leap before you look”,

“judgment-oriented” vs. “action-oriented”, or “wait and see”/ ”watchful waiting” vs. action.


1982) and it can be seen as the rational or “proper” sequence to problem solving. In

general, the rational decision making approach consists of the stages problem definition,

diagnosis, design of alternatives, and selection of the preferred option (Mintzberg &

Westley, 2001). These steps are conducted sequentially, with diagnosis as an action-

enabling step. As organizational decision making processes tend to be incremental, they

involve iterative cycles of design and readjustment (Mintzberg et al., 1976).

Ready-fire-aim (RFA). In practice, firms often do not follow the rational RAF

strategy. Empirical insights reveal that managers are inclined to prefer quick action to

analysis (Isenberg, 1984; Mintzberg et al., 1976). The studies of Mintzberg (1973) about

what managers actually do, as opposed to what they are supposed to do or what they say

they do, indicate that rational problem solving was rarely employed. Instead, an action-

orientation has frequently been observed (Isenberg, 1986; Mintzberg & Westley, 2001).

Taking action provides orientation and information that enables to learn and adjust

response efforts (Rudolph, Morrison, & Carroll, 2009). We call this strategy ready-fire-

aim. It has been recognized in innovative firms, which tend to act rather than analyze,

plan, and postpone action (Peters & Waterman, 1982). Weick (1979) argued that

“postponing action while planning continues could prove dangerous [...] and any chance

of clarifying the situation will decrease, simply because there is nothing available to be

clarified or made meaningful” (p. 103). In case of dramatic change, he suggested that

“chaotic action is preferable to orderly inaction” (p. 245). Daft and Weick (1984) added

that “feedback from organizational actions may provide new collective insights” (p. 286)

which may improve both the understanding of a situation as well as the ability to

implement an effective response. Accordingly, RFA is based on Weick’s (1979) idea of

organizations engaging in iterative cycles of action and reaction to gradually reduce

uncertainty related to environmental information.

In our model, a firm that applies RAF uses the first period after a supply disruption

to collect information and resolve the response uncertainty while its choice configuration

remains unchanged. Afterwards, the firm engages in local search and is able to perfectly

predict the fitness values of alternative configurations. In contrast, a firm that applies

RFA immediately starts local search after a supply disruption and faces noisy information

due to response uncertainty. In line with G. Wu (1999) – who suggested that “uncertainty

about the outcome of a particular state of nature is often not resolved immediately after

an act is selected” (p. 159) – we assume that response uncertainty is dynamic and does


not completely vanish after action has been taken. Hence, the fitness values of both a

firm’s current configuration and alternative configurations are distorted by uniformly

distributed random errors for D periods. The random errors are newly generated in each

period and are part of an interval of a pre-defined value E of standard deviations (SD) of

all fitness values of a given environment and range from –E × SD to E × SD. The size of

this interval decreases linearly over time. If the duration of response uncertainty D equals

5, the interval will shrink by 20 percent of its initial size in each period and the response

uncertainty is resolved after 5 periods.

Figure 8 depicts an individual simulation run over 80 periods of time. For every

period of time (x-axis), the corresponding fitness value for each firm (agent) is plotted on

the y-axis. The firm pursuing RFA immediately responds to the disruption in period 40

although fitness values of choice configurations are distorted (FRFA, perceived, dotted black

line). However, the true fitness value (FRFA, true, solid black line) is reduced by some

alterations. The firm pursuing RAF acts later, but does not experience detrimental

decisions during its recovery (FRAF, solid gray line). Finally, both firms recover from the

disruption’s negative consequences reaching different long-term fitness values.

Figure 8: An individual simulation run

Note. The simulation run was conducted in a setting with N = 6, Khigh, Emedium, Dlong, and PDnone.


4.3.6 Path dependence

An important characteristic of actual organizational adaptation processes is path

dependence (Beinhocker, 1999). Path dependence means that “where we go next depends

not only on where we are now, but also upon where we have been” (Liebowitz &

Margolis, 2000, p. 981). Adaptation decisions may influence the available range of future

options and future flexibility (Nelson, Adger, & Brown, 2007). Serious problems may

occur if actions that cannot be reversed in the short term prove to be a “shadow of the

past” which results in inferior outcomes creating lock-in situation (J.-P. Vergne &

Durand, 2011). Hence, in the face of environmental change, path dependence can force

firms on a suboptimal course of action (Noda & Collis, 2001). Recent research has

emphasized the need for a better understanding of path dependence (e.g., Sydow,

Schreyögg, & Koch, 2009; J. P. Vergne & Durand, 2010).

Recovery from supply disruptions may involve supply chain reconfigurations that

are partly or completely irreversible. Moreover, organizational inertia can prevent firms

from making possible changes. Imagine that one choice of a firm’s decision set concerns

a supplier selection decision. It is unrealistic to assume that the decision for a specific

supplier can immediately be reversed in the following period without excessive switching

costs and time. Against this backdrop, we adopt the view of path-dependent

organizational action in the context of adaptation to environmental change. To this end,

we include the parameter PD in our model. PD determines the number of a firm’s first

alterations that are irreversible. If PD equals 0, all decisions can be reversed in subsequent

periods. However, if, for example, PD equals 2, this means that the first two conducted

changes cannot be reversed again and keep their state for the rest of the simulation. Due

to this path dependency through irreversible actions, firms can find themselves “locked-

in” and not reaching any peak of an environment.

In sum, our model extends the standard NK model as used in previous research. We

include response uncertainty in the evaluation of fitness values by means of the

parameters E and D, and incorporate irreversible actions via the parameter PD. Thus, the

model allows the investigation of response uncertainty, complexity, and path dependence

in the context of disruption recovery.


4.3.7 Results

The goal of our simulation experiments is to analyze the conditions under which a firm

should decide to delay the initial recovery decision to a point where more accurate

information has been accumulated. We generate two firms (agents) with identical initial

configurations d (without loss of generality) at the beginning of each individual

simulation run of 80 periods of time (T). Both firms face six decisions (N = 6) and operate

in the same environment with either low (Klow = 1) or high complexity (Khigh = 5). This

environment is stable until it is disrupted in period 40. The disruption results in an

environmental change and the firms’ environment is replaced by a randomly generated

new one. The parameters N and K are not altered by the disruption. Subsequently, one

firm applies purely RAF to eliminate response uncertainty before it takes action, while

the other applies purely RFA facing distorted fitness values to a pre-defined degree E for

a certain number of time periods D. Furthermore, a certain number of the firms’ first

alterations PD is irreversible.

By varying the four parameters K, D, E, and PD, the simulation model enables

sophisticated analyses under a variety of conditions. We focus on parameter values that

create diverse settings to derive the implications of RAF and RFA for the recovery

performance of firms. The parameter values used in the simulation experiments are

depicted in Table 11.

Given this setup, we compare the average recovery performance of RAF and RFA

in different settings of our simulation experiments based on four measures which address

two main dimensions: Effectiveness and speed. The performance measures we report

constitute averages of 10,000 runs to eliminate the stochastic component that any single

run is sensitive to. 8

Effectiveness. The effectiveness of a firm’s recovery efforts is a relevant

performance indicator, because it reflects the extent to which negative effects on firm

performance have been minimized (Bode et al., 2011; Handfield et al., 2007). We

8 Two additional performance measures – “average first improvement” and “alterations required to recover”

– were scrutinized, but did not reveal interesting results beyond the rather obvious insight that RAF always

leads to more precise initial actions and requires less changes to recover from disruptions than RFA. For

the rest of this paper, we focus on measures that identify different dominant strategies depending on the

environmental conditions. Since the degree of complexity influences the absolute levels of fitness in NK

landscapes, we measure fitness values in relation to the global optimum of an environment (Skellett,

Cairns, Geard, Tonkes, & Wiles, 2005).


construct two effectiveness-related performance measures: Post-disruption performance

(PDP) and final performance (FP).

Speed. Anecdotal evidence suggests that the speed of completing the recovery

reduces costs and losses. For example, Macdonald and Corsi (2013) argued that “the

longer it takes to fully recover, the more expensive the entire recovery process is likely

to be” (p. 272). Two measures are used to capture speed: Uncertainty resolution

performance (URP) and number of time periods (NTP). Whenever we report that one

response strategy (RAF/ RFA) performs better than the other, the difference in mean

performance is statistically significant with p < 0.05 or better.

Table 11: Parameter values determining the experimental conditions

(1) Number of decisions (N), one value

(i) N = 6

(2) Complexity (K), two values

(i) Low: Klow = 1

(ii) High: Khigh = 5

(3) Duration of response uncertainty (D), in time periods, two values

(i) Short: Dshort = 5

(ii) Long: Dlong = 10

(4) Response uncertainty (E), in standard deviations of an environment’s fitness values, three values

(i) Low: Elow = 0.5

(ii) Medium: Emedium = 1

(iii) High: Ehigh = 2

(5) Path dependence (PD), in decisions, two values

(i) None: PDnone = 0

(ii) Strong: PDstrong = 4

Figure 9 summarizes the conditions under which a certain response strategy

dominates the other on a given performance dimension (effectiveness or speed).

Managers should be aware that RAF is more precise in the first period of search and

requires less alteration and related investment to find a long-term configuration than RFA

in all depicted settings. However, from interviews with procurement executives that we

conducted as part of this study, the perception that, in case of a supply disruption, costs

are regarded as “necessary evil” became apparent, because the goal of a recovery process

is “to maintain the supply of customers, whatever the cost.”

Post-Disruption Performance (PDP). In our simulation, the PDP is the average

fitness value of a firm from period 40 to the end of the simulation run. It demonstrates the

extent to which negative disruption effects have been minimized. Figure 10 depicts the

PDPs of RAF and RFA under various conditions.


Figure 9: Summary of results

Note. The results at the bottom of the tree indicate which strategy (RAF or RFA) is superior in terms of the performance dimensions effectiveness (PDP and FP) and

speed (URP and NTP) given the specific parameter configurations. A strategy is superior in a given performance dimension if at least one of both KPIs (PDP and FP for

effectiveness, URP and NTP for speed) reflects a better performance while the strategy does not perform weaker in the other KPI. Blank cells (white background, filled

with “–“) indicate that a superior strategy cannot be determined based on the simulation experiments. (a) “no” refers to no path dependence (PDnone), “stg” refers to strong path dependence (PDstrong).

Effectiveness

Speed

Complexity

(K)

Response

uncertainty

(E)

Duration (D)

Shadow of

the past

(SOP)a

low

no stg no stg

low

short long

no stg no stg

medium

short long

no stg no stg

high

short long

no stg no stg

low

short long

no stg no stg

medium

short long

no stg no stg

high

short long

high

Supply disruption recovery

RFA – RFA RAF RFA RAF RFA RAF RFA RAF RFA RAF RFA RFA RFA RFA RFA RFA RFA RFA RFA RAF RFA RAF

RFA RFA – RFA RAF RFA RAF RAF RAF – RAF RAF RFA RFA – RFA RFA RFA – RFA RAF RAF RAF RAFResu

lts

Pa

ram

ete

rs


Figure 10: Post-disruption performance (PDP)

The complexity of the respective environments is indicated as either low (Klow = 1)

or high (Khigh = 5). The diagrams’ x-axes depict three levels of response uncertainty: Low

(Elow = 0.5), medium (Emedium = 1), and high (Ehigh = 2). Disruptions can either be followed

by a short (Dshort = 5) or long (Dlong = 10) period of uncertainty and decisions are either

reversible (PDnone = 0) or the first four altered decisions are irreversible (PDstrong = 4). For

each possible combination of D and PD, one of the four diagrams shown in Figure 10

depicts the average PDP of RAF and RFA on the y-axis. Every diagram comprises four

lines. The dotted lines refer to RFA and the solid lines to RAF. This basic structure for


the presentation of the simulation results remains the same for all figures which are used

in the results section hereafter.

The results shown in Figure 10 provide several interesting insights. If response

uncertainty is low, RFA dominates RAF in all settings with respect to PDP. Furthermore,

in highly complex environments, this performance advantage of RFA extends to medium

response uncertainty. In highly complex environments where actions are reversible, RFA

outperforms RAF, regardless of the level and duration of response uncertainty (D).

However, the advantage of RFA is reduced by increasing path dependence (PD). In highly

complex environments where actions are irreversible, strong path dependence

accompanied by high response uncertainty results in RAF achieving a greater or at least

equal PDP compared to RFA. In environments with low complexity and high response

uncertainty, RAF dominates RFA, regardless of the intensity of path dependence. If we

focus on the implications for complex landscapes (given the premise that most firms

operate in complex supply chains and environments), these insights lead to the following

propositions:

Proposition 1. If the environment is highly complex and actions are reversible, then

PDPRFA > PDPRAF.

Proposition 2. If the environment is highly complex, actions are irreversible, and

response uncertainty is low or medium, then PDPRFA > PDPRAF.

Final Performance (FP). The aim of disruption response is to restore desired

performance levels and a firm’s ability to do this is depicted by the fitness level of its

long-term configuration (FP). To this end, we examine the final fitness values (period 80)

of RAF and RFA. Figure 11 reveals that in environments with low complexity,

irreversibility of actions leads to greater or at least equal final fitness values for RAF

compared to RFA. RAF’s performance advantage increases with the level of uncertainty.

However, if actions are reversible, RFA leads to greater or at least equal final fitness

values compared to RAF in all settings. In highly complex environments, this advantage

of RFA increases with the level and duration of response uncertainty. However, if

decisions in highly complex environments are irreversible, RAF outperforms RFA if

response uncertainty is high. Hence:

Proposition 3. If actions are reversible, then FPRFA ≥ FPRAF.

Proposition 4. If the environment is highly complex and actions are reversible, then

FPRFA increases with the level and the duration of response

uncertainty.


Proposition 5. If the environment is highly complex, actions are irreversible, and

response uncertainty is high, then FPRAF > FPRFA.

Figure 11: Final performance (FP)

Uncertainty Resolution Performance (URP). URP depicts how quickly firms have

been able to minimize the negative effects of a disruption on their performance (fitness

value) at the point of time where response uncertainty has completely been resolved (either

period 46 or 51). From then on, the firms face undistorted fitness values.

Figure 12 shows the URPs observed in our simulation experiments. One of the main

insights from Figure 12 is that in highly complex environments RFA leads to a greater

(or at least equal) URP than RAF if response uncertainty is low or medium. In highly


complex environments with high and long-lasting response uncertainty, RAF leads to a

greater URP than RFA. When response uncertainty is low or medium and short-lived,

RFA achieves a better URP than RAF, independent of the complexity of the firms’

environment or the intensity of path dependence. Accordingly, we propose the following:

Proposition 6. If the environment is highly complex and response uncertainty is low

or medium, then URPRFA ≥ URPRAF.

Figure 12: Uncertainty resolution performance (URP)

Number of Time Periods Required to Reach the Long-Term Configuration

(NTP). As a further measure of recovery process speed, we investigate how quickly firms

recover.


Figure 13: Number of time periods required to reach long-term configuration (NTP)

To this end, the average number of time periods needed to reach a stable long-term

configuration is calculated. Figure 13 depicts the NTP that firms applying one of the

response strategies require to find their long-term configuration, subsequent to a supply

disruption. In settings where actions are reversible and response uncertainty is long-

lasting, RAF will be faster than (or at least as fast) RFA in reaching the long-term

solution. However, if actions are irreversible, RFA leads to a more quickly completed

recovery if response uncertainty is only low or medium and short-lived. This leads to the

following and last propositions:


Proposition 7. If actions are irreversible and response uncertainty is neither high

nor long-lasting, then NTPRFA < NTPRAF.

Proposition 8. If actions are reversible and response uncertainty is medium or high,

then NTPRAF < NTPRFA.

4.3.8 Robustness of simulation results

To ensure that the results obtained were not due to chance, we performed several checks

in congruence with previous studies using the NK model. First, the representativeness of

the results was ensured by conducting 10,000 replications for every combination of K, D,

E, and PD to eliminate the influence of random fluctuations in the generation of firms

and environments. Every time we report a performance difference between the two

response strategies, this difference in means is statistically significant with p < 0.05 or

better. Second, we systematically varied all of the model parameters. To be as clear and

concise as possible, we have presented only a subset of the results of all these simulation

experiments. We focused on polar types of parameter values to be able to identify a

parameter’s influence. Although the precise quantitative outcomes change, the qualitative

patterns can also be observed with intermediate values of the model parameters K, D, E,

and PD. We find that most of the results that we report for highly complex environments

(see P1, P4, P5, and P6) also hold if K equals 4. All other results hold for all values of K.

Furthermore, the results we report for settings with irreversible actions (see P2, P5, and

P7) hold for values of PD > 0. Increasing the number of decisions N does not qualitatively

change the results. Third, the results are not sensitive to the assumption of similar initial

configurations for the two agents in every simulation run. The same qualitative patterns

also result from unequal initial configurations.

4.3.9 Discussion

By means of simulation experiments, we investigated under which conditions managers

should delay their response actions until they have acquired more evidence on the

consequences of response actions. The propositions derived are robust in many regards

and yield several general insights. In the following, we focus on the insights generated

for highly complex environments. We assume them to resemble today’s business

environments more accurately than less complex environments.

First, an immediate response to supply disruptions is beneficial for the average post-

disruption performance even if the consequences of response actions cannot be


determined precisely. This holds true even if more than half of the decisions are

irreversible and response uncertainty is long-lasting. Only if the latter becomes very high

so that managers are virtually “poking around in the dark”, firms should delay action until

further information has been collected.

Second, the choice of a response strategy significantly affects the average long-term

performance of firms. Moreover, if actions are reversible, firms pursuing RFA benefit

from a high level of response uncertainty. This seems to be quite counterintuitive.

However, one explanation for this result might be that high levels of response uncertainty

drive exploration and testing of configurations which may turn out to be detrimental in

the short-run but enable firms to achieve superior performance in the long-run. Hence,

early decisions with an adverse effect on a firm’s performance can provide an opportunity

for future local improvements if actions are reversible. If detrimental initial decisions

cannot be reversed, they pose a burden that a firm cannot recover from.

Third, although immediate action subsequent to a supply disruption requires more

changes to find a stable long-term configuration, firms recover more quickly if response

uncertainty is neither high nor long-lasting. This means that if firms are capable of quickly

improving their ability to predict the consequences of their response actions, they benefit

from a quick reaction and recover faster than competitors that might delay action to spend

further resources on information collection.

4.4 How managers actually intend to behave in supply

disruption response situations

To complement the findings from the simulation experiments, we use a vignette-based

experimental methodology to analyze the behavior of managers in disruption recovery

processes in business-to-business (B2B) contexts. Vignette-based experiments offer

several methodological advantages which fit our specific research context. First, they

provide controlled tests of causal relationships by carefully manipulating vignettes that

are presented to the participants. Second, in comparison to rather retrospective studies

based on surveys or case studies, the use of vignette-based experiments can generate more

reliable data on respondent behavior, because participants have to indicate their intentions

shortly after reading a specific scenario to minimize retrospective bias (Wathne et al.,

2001). Third, an experimental design enables researchers to study behavior and choices

where individuals or firms are usually not likely to share information (Rungtusanatham


et al., 2011). A supply chain disruption, which is our studied context, is typically

accompanied by adverse effects on firms. Furthermore, severe incidents are rather rare

and it would be unethical to actually disrupt a firm to collect data. Vignette-based

experiments help to overcome both issues.

A vignette is defined as “short, carefully constructed description of a person, object,

or situation, representing a systematic combination of characteristics” (Atzmüller &

Steiner, 2010, p. 128). In our experiment, we integrate vignettes into a survey, since this

is a promising but rarely applied approach to study respondents’ judgments and account

for each approach’s shortcoming. By randomly assigning participants to treatment cells

we eliminate systematic differences in the participants that might affect their responses

(e.g., Bachrach & Bendoly, 2011). Hence, differences in response behavior can be

attributed to the manipulated experimental treatments. This methodological approach has

recently been employed in the field of operations management to study “chain liability”

(Hartmann & Moeller, 2014), make-or-buy decisions (Mantel et al., 2006), observational

learning (Hora & Klassen, 2013), and perceptual differences between buyers and

suppliers (Ro et al., 2016).

4.4.1 Development of the vignettes

A key aspect of vignette-based experiments is the vignette design and validation. We

carefully constructed vignettes that assigned participants to the role of a procurement

manager that faces a supply disruption and considers to take immediate action to mitigate

its negative consequences. We followed the principle of form postponement, hence, all

vignettes began with an introductory common module that was the same for all

participants to ensure the provision of a similar contextual background (Aguinis &

Bradley, 2014; Rungtusanatham et al., 2011). Our three factors of interest, response

uncertainty, complexity, and path dependence, were manipulated in a subsequent

experimental cues module. We developed a scenario of a supply disruption involving a

supplier that is unable to ramp up production and deliver large enough amounts of its

product (electric engines) as an example of a disruption that frequently occurs in practice

(e.g., Hepher & Altmeyer, 2017; D. Wu & Ting-Fang, 2017). All other factors of the

vignettes were held constant.

Our factors of interest were manipulated in line with the parameters and their levels

shaping the model developed in the third section to be able to depict comparable


scenarios. To maintain an adequate degree of complexity for the participants, we

restricted the experiment to a static one-shot decision, focused on low and high levels of

response uncertainty, and abstained from including duration of response uncertainty as a

further factor. In total, using a 2 (response uncertainty: Low or high) x 2 (complexity:

Low or high) x 2 (path dependence: None or strong) full factorial design, we created eight

vignettes. After reading a vignette, participants were asked to imagine themselves in the

role of the procurement manager in the described scenario and indicate their willingness

to take immediate action. We utilized a form of indirect questioning (projective

technique) to limit potential demand characteristics and effects of social desirability

(Thomas et al., 2013). In addition, respondents were supposed to report their risk attitude.

In order to measure the latter, we employed the general risk question (Dohmen et al.,

2011) which is widely used and requires participants to rate their individual willingness

to take risks on a 11-point Likert-type scale. It has been shown to be a very good predictor

of risk-related behavior.

Two different levels of response uncertainty were described by varying how

uncertain it was that the proposed action positively influenced the depicted supply

disruption’s negative consequences. It was either quite certain (low) or very uncertain

(high) that the consequences of taking action could be predicted. Complexity was varied

by manipulating the independence of the decision. In the low complexity setting, the

decision needed to be taken in a slightly complex environment with minor impact on other

internal processes and its success was almost independent of the actions of colleagues. In

the high complexity setting, the decision needed to be taken in a very complex

environment with a very large impact on other internal processes and its success was very

much dependent on the actions of colleagues. Path dependence was manipulated by

varying the extent to which future options were limited by the decision to immediately

act or not. They were either not (none) or heavily (strong) limited.

As recommended by Wason et al. (2002), the vignettes were pre-tested with 85

students that were randomly assigned to one of the eight treatment condition to assess the

vignettes’ validity, internal consistency, and realism. After reading the vignettes, students

were asked to indicate their willingness to act and responded to manipulation checks to

ensure that different levels of the independent variables were appropriately represented.

Average responses for complexity and path dependence were significantly different

between low and high treatment groups. We included three levels of response uncertainty


in our pretest (Low, medium, and high), however, there were no significant differences

between medium and high response uncertainty. In order to address this shortcoming and

reduce the complexity for our participants, we focused on low and high levels of response

uncertainty in the final experiment. The students also rated the realism for the scenario

description on a seven-point scale (1 = “not at all” to 7 = “totally”). The vignettes

appeared realistic with mean responses ranging from 4.50 to 6.29 (grand mean is 5.61).


Data were collected by means of a self-administered online experiment. Between June

and September 2017, 1056 managers with direct experience in supply chain management

from Germany, Austria, and Switzerland, have been invited to participate in our

experiment. Contact addresses were obtained from a commercial business data provider.

The managers received an invitation via e-mail and were randomly assigned to one of the

eight treatment conditions. 112 of them completed the experiment, resulting in an

effective response rate of 10.61%. On average, participants had almost 18 years of

experience in supply chain management (SD = 10.33). Two observations were dropped

due to outlier analysis, resulting in 110 usable vignettes.

4.4.3 Dependent variable

By means of our vignette experiments, we aim to reveal how response uncertainty,

complexity, and path dependence affect the intentions of managers to take immediate

action in response to supply disruptions. The decision to take immediate action can be

seen as being binary, however, binary outcomes provide only little variance to explain

and understand the effects of response uncertainty, complexity, and path dependence on

such decisions. We follow McKelvie et al. (2011) and Cantor, Blackhurst, and Cortes

(2014) by operationalizing our dependent variable as the participant’s intention to take

immediate action (ITA), on a nine-point scale (anchored at 1 := “not at all likely” to 9 :=

“totally likely”). The specific action to be considered is described in detail in each

vignette.

4.4.4 Manipulation checks

As in the pre-test, we verified that each participant perceived differences in our

manipulations of the independent variables, as intended. To this end, all participants were


asked to respond to several manipulation checks, on seven-point scales (1 := “not at all”

to 7 := “totally”), after reading and responding to a vignette. To assess the perceived level

of response uncertainty, participants were asked to indicate to which extent they agree

that the consequences of the proposed action were very uncertain. The respondents’

perception of low and high levels of response uncertainty were significantly different

(Mlow = 2.61, Mhigh = 5.92; t(108), p < 0.001). To validate the manipulation of complexity,

we asked participants to evaluate whether the decision had to be taken in a very complex

environment. The participants’ responses were significantly different for low and high

levels of complexity (Mlow = 1.85, Mhigh = 6.04; t(108), p < 0.001). Finally, to check

whether the two levels of path dependence were perceived to be different, participants

rated to which extent they agree that the depicted decision limited future opportunities to

a very large extent. Respondents reported higher degrees of agreement when path

dependence was strong compared to when it was absent (Mnone = 2.17, Mstrong = 5.49;

t(108), p < 0.001).

4.4.5 Results

We relied on analysis of variance (ANOVA) to dissect our participants’ intentions.

Parametric tests (e.g., ANOVA) are appropriate and robust for Likert-type data and can

be used with “small sample sizes, with unequal variances, and with non-normal

distributions, with no fear of ‘coming to the wrong conclusion’” (Norman, 2010, p. 631).

A three-way ANOVA with response uncertainty (low or high), complexity (low or high),

and path dependence (none or strong) as between-subjects factors and ITA as dependent

variable was conducted. Table 12 contains information on the number of participants per

cell, cell means, and corresponding standard deviations for our dependent variable ITA.

The results revealed significant main effects of response uncertainty (F = 25.07, p

< 0.001), complexity (F = 17.19, p < 0.001), and path dependence (F = 4.77, p = 0.03) on

ITA. Low response uncertainty led to higher levels of ITA (Mlow = 5.02, SD = 2.60) than

high response uncertainty (Mhigh = 2.94, SD = 1.91). The low complexity treatment

revealed higher levels of ITA (Mlow = 5.00, SD = 2.46) than the high complexity treatment

(Mhigh = 3.14, SD = 2.24). Finally, the absence of path dependence resulted in a higher

willingness to take immediate action (Mnone = 4.58, SD = 2.53) than its presence (Mstrong

= 3.56, SD = 2.43). No statistically significant interaction effects between these three


variables were identified. ANOVA results are provided in Table 13 and Figure 14 depicts

the analyzed main effects.

Table 12: Frequencies, cell means, and standard distributions for ITA

Experimental condition Intention to take

immediate action (ITA) Number of

observations

(n = 110) Path

dependence Complexity

Response

uncertainty M SD

None

Low Low 6.88 1.75 16

High 3.92 2.02 12

High Low 4.33 2.61 12

High 2.62 1.50 13

Strong

Low Low 5.29 2.09 14

High 3.25 2.42 12

High Low 3.53 2.67 17

High 2.14 1.35 14

Table 13: ANOVA results


Response uncertainty 110.96 1 110.96 25.07 0.000 ***

Complexity 76.09 1 76.09 17.19 0.000 ***

Path dependence 21.11 1 21.11 4.77 0.031 *

Response uncertainty × Complexity 6.04 1 6.04 1.36 0.246

Response uncertainty × Path dependence 2.66 1 2.66 0.60 0.440

Complexity × Path dependence 1.62 1 1.62 0.37 0.546

Response uncertainty × Complexity × Path dependence 0.59 1 0.59 0.13 0.716

Model 240.21 7 34.32 7.75 0.000 ***

Residual 451.47 102 4.43

Note. n = 110. The dependent variable is “intention to take immediate action” (ITA). SS refers to “sum of

squares” (Type III), df to “degrees of freedom”, and MS to “mean square”.

* p < 0.05, ** p < 0.01, *** p < 0.001.

To examine the robustness of our findings and to ensure that the participants’

intentions were not determined by their experience and risk attitude, we included both

variables as covariates in an analysis of covariance (ANCOVA) with ITA as dependent

variable and response uncertainty, complexity, and path dependence as independent

variables. Neither the experience nor the risk attitude of the participating individuals had

a statistically significant influence on ITA. Hence, the predicted effects remained

qualitatively similar.


Figure 14: The effects of response uncertainty, complexity, and path dependence on

ITA

4.4.6 Discussion

The depicted results provide insights into the intentions of managers to take immediate

action after being hit by a supply disruption. The carefully designed vignette experiment

reveals that the main parameters of our supply disruption recovery model, namely

response uncertainty, complexity, and path dependence, considerably affect actual

decision making within disruption management. Furthermore, we find that the effects of

these influencing factors on managers’ willingness to take immediate action do not

interact with each other which enables straightforward interpretation.

More specifically, we demonstrate that, as response uncertainty and path

dependence increase, managers are less willing to take immediate action. This means that,

with regard to these two influencing factors, managers’ intentions basically correspond

with how they should behave according to the simulation experiments. However, the

results also demonstrate that higher levels of complexity seem to inhibit quick responses,

which contradicts the insights drawn from the simulation experiments. The latter suggest

that, as complexity increases, immediate action becomes more beneficial and should be

preferred to delaying action if response uncertainty is not high. In contrast, our

participants generally reveal lower levels of their willingness to take immediate action.

This is in line with prior research, which states that decision makers will prefer to delay

action, seek new alternatives, or revert to the status quo as the choice environment


becomes complex (Dhar, 1997). However, we thereby identify and provide empirical

support for an important mismatch between how managers should behave and how they

actually intend to act.

4.5 General discussion

Recovering from supply disruptions is a challenging task for the responsible managers.

Typically, the subsequent decision making process is characterized by response

uncertainty, complexity, and path dependence. Response uncertainty results from a lack

of being able to precisely predict the consequences of response alternatives, complexity

is driven by interactions among decisions, and path dependence stems from considerable

expenditures required to change the operational setting of a firm. Against this backdrop,

our multi-method study contributes to the emerging literature on supply disruption

management by (1) providing insights into disruption recovery processes, (2) clearly

defining RAF and RFA and delineating their implications for the recovery performance

of firms by providing testable propositions, (3) revealing actual intentions of managers

regarding their willingness to take immediate action in response to supply disruptions,

and (4) considering distorted performance evaluations in the NK model by incorporating

uncertainty.

4.5.1 Implications for research

This research analyzes the conditions under which a firm should decide to delay the initial

recovery action to a point where more accurate information has been accumulated and

investigates how managers actually intend to act. In this context, we have identified two

archetypical approaches of organizational decision making: RAF and RFA. The former

characterizes firms that tend to collect further information before taking action in the

hopes of being able to make better informed decisions by eliminating response

uncertainty. This means that analysis precedes and enables action. Thus, RAF resembles

the rather rational approach to problem solving, similar to what one of the procurement

executives that we interviewed explained: “Discovery, then we collected information.

[…] Then you consider, what can you do? You develop options and alternatives.

Afterwards you try to decide which one can be applied. Then, implementation and

monitoring.”


RFA represents firms that respond immediately to a supply disruption, even if

response uncertainty is present. Procurement executives stated the following when asked

about their response to supply disruptions: “We immediately started bottleneck-

management, before we exactly knew how the situation really is”, “acting started

immediately after the discovery”, and that they “did not initially build large teams to

discuss it, we rather needed to act very quickly.” Based on imprecise or incomplete

information on the consequences of response alternatives, they took action to be able to

observe results that enable further adjustment. One respondent described such a process:

“We tried to find short-term solutions first. We took action and when we […] did not

reach the performance that was supposed to be reached, we decided to take another

path.” Following Weick’s (1979) idea, these firms emphasize the use of action to make

sense of observable outcomes. Thus, action is assumed to clarify the situation and create

meaningful insights to guide and enable subsequent decisions.

Table 14: Summary of propositions

P1 If the environment is highly complex and actions are reversible, then PDPRFA > PDPRAF.

P2 If the environment is highly complex, actions are irreversible, and response uncertainty is low or

medium, then PDPRFA > PDPRAF.

P3 If actions are reversible, then FPRFA ≥ FPRAF.

P4 If the environment is highly complex and actions are reversible, then FPRFA increases with the

level and the duration of response uncertainty.

P5 If the environment is highly complex, actions are irreversible, and response uncertainty is high,

then FPRAF > FPRFA.

P6 If the environment is highly complex and response uncertainty is low or medium, then URPRFA ≥

URPRAF.

P7 If actions are irreversible and response uncertainty is neither high nor long-lasting, then NTPRFA

< NTPRAF.

P8 If actions are reversible and response uncertainty is medium or high, then NTPRAF < NTPRFA.

Different authors have advocated either RAF (e.g., Kepner & Tregoe, 1965) or RFA

(e.g., Peters & Waterman, 1982), but the implications of both approaches on a firm’s

disruption recovery performance in complex settings accompanied by response

uncertainty and path dependence have not yet been examined. The findings of our

simulation experiments, summarized in eight testable propositions shown in Table 14,

address this important research gap. Obviously, firms that apply RFA are exposed to the

risk of erroneously altering their operating procedures in the belief that the change will

lead to better firm performance. Although the expected performance of an alternative

configuration might be greater than the performance of the current configuration, the true

performance might be reduced, because the performance values of alternative


configurations are distorted by response uncertainty in the early stages of a recovery

process.

In line with the insights provided in the third section, we furthermore demonstrate

that response uncertainty and path dependence reduce the actual willingness of managers

to take immediate action. However, higher levels of complexity have a similar effect,

although managers should actually respond quickly to supply disruptions in complex

environment if response uncertainty is not high. This circumstance illustrates a

considerable mismatch between managers’ intentions and their optimal behavior

according to our findings. Moreover, we do not find evidence for an action bias in post

supply disruption decision making (Bar-Eli, Azar, Ritov, Keidar-Levin, & Schein, 2007).

On the contrary, managers react sensitively to high levels of response uncertainty,

complexity, and path dependence and reduce their willingness to take immediate action.

4.5.2 Implications for practice

In addition to the theoretical implications discussed above, the results of our research also

have important implications for practice. Managers responding to supply disruptions need

to be aware of the conditions that characterize their environment. The level of response

uncertainty that firms face considerably affects the ramifications of delaying action or

immediately responding to a supply disruption. In complex environments, only if

response uncertainty is perceived to be high and long-lasting, waiting for more

information can be advisable, based on our results. Furthermore, managers are required

to trade off the accuracy of their actions against the speed of their reaction. On the one

hand, delaying a response results in more precise first improvements and requires less

changes to be made. On the other hand, immediate responses on average result in quicker

recovery of firm performance. Moreover, the long-term performance of firms is, on

average, not weakened but rather strengthened by response uncertainty if detrimental

decisions can be reversed. In complex environments, firms may also benefit from quick

responses in the short run and may increase their market share as Nokia did in did in the

aftermath of the aforementioned disruption.

Furthermore, previous research on decision making behavior highlighted the

influence of cognitive biases on the decision to act or not to act. Managers should be

aware of these biases to avoid being misled in their decision of when to respond to a

supply disruption. Some decision makers tend to be biased towards analysis (Kerstholt,


1996). A tendency for RAF can, for example, even be fostered by a status-quo bias

(Samuelson & Zeckhauser, 1988). This bias represents the tendency in human decision

making under uncertainty to value the current state more than a potentially superior but

uncertain alternative. In the worst case, this may lead to the phenomenon of analysis

paralysis meaning that action is delayed further and further. Similarly, a tendency for

RFA is intensified by an action bias. If managers take action, “at least they will be able

to say that they tried to do something” (Bar-Eli et al., 2007, p. 616). Furthermore, action

is considered more appropriate than inaction in response to bad performance (Zeelenberg,

Van den Bos, Van Dijk, & Pieters, 2002). Hence, taking action might often appear to be

an attractive option for managers although they should rather delay a response.

Surprisingly, based on the results of the vignette-based experiments, we are able to

demonstrate that managers tend to refrain from taking immediate action if the degree of

either response uncertainty, complexity, or path dependence increase. However,

managers should rather take immediate action in complex environments, as delineated by

our simulation experiments. This incongruity has important practical implications for the

management of supply disruptions. Managers seem not to be aware of the benefits

associated with quick responses in complex decision making environments and should

demonstrate a higher willingness to take immediate action when exposed to complexity.


The contribution of this research is constrained by several limitations. As a main

assumption of the NK model and the model developed in the third section, each decision

i is assumed to interact with exactly the same number of decisions as every other decision.

Nevertheless, some decisions might actually interact with more of the other decisions than

others do. Furthermore, we have limited our model to decision makers with centralized

authority, because we assumed that this is a basic characteristic of disruption management

processes. We did not consider many factors that might also characterize organizational

search processes such as the presence of a hierarchy (Mihm et al., 2010) and the need to

coordinate (Lounamaa & March, 1987). In addition, we assumed that a supply disruption

does not change the underlying complexity of an environment. However, it might be

possible that a severe supply disruption alters the level of interaction between the

operational activities of a firm. Another limitation concerns our conceptualization of

performance. Although we refer to a firm’s operating performance as main determinant


of a firm’s recovery efforts, our model’s measure of operating performance is abstract

rather than specific.

Moreover, although complementing the simulation experiments, the vignette-based

experiments introduces further limitations. First and foremost, the participants were

exposed to vignettes containing information about a specific one-shot supply disruption

situation. In real disruption recovery processes, subjects are likely to receive such

information in a more fragmented fashion, perhaps over multiple time periods. In order

to reduce the complexity of the decision making task for our respondents, we refrained

from a multi-stage setting taking, e.g., the duration of response uncertainty into account.

Thus, research accounting for the dynamics of disruption recovery processes would add

to the validity and generalizability of our findings. In addition, we rely on the intentions

of managers instead of their actual behavior. However, prior research shows that

intentions may serve as reliable indicators of actual behavior (e.g., Ajzen, 1991; T. L.

Webb & Sheeran, 2006).

Finally, our work suggests several opportunities for further research. The delineated

and defined disruption response strategies RAF and RFA could be empirically

investigated through large-scale studies or further (dynamic) experiments. The developed

propositions could be used to derive testable hypotheses on the performance of RAF and

RFA. Moreover, our research indicates that the use of the NK model can provide rich

insights into the management of supply disruptions. The NK framework will remain a

powerful tool to analyze organizational decision making under complexity that does not

allow for analytical optimization. Future research could build on our suggestions on how

to represent supply disruption recovery and apply the NK model methodology to further

research questions of the field by adjusting or extending our model. In addition, our

research demonstrates the importance and relevance of behavioral aspects in the supply

(chain) disruption context and the need to further investigate disruption management

processes, typically characterized by limited time and unreliable or sparse information.

Therefore, we hope that this work encourages further research on supply disruption

management. Given the fact that firms will most likely never be able to fully control their

environment and perfectly predict changes, managers will continue to face severe supply

disruptions and require additional insights on how best to respond to them.

Conclusion and future research directions 94

Chapter 5 Conclusion and future research

directions

This chapter summarizes the research delineated in the previous chapters and highlights

the main answers to the research questions formulated in Chapter 1. In addition, the main

limitations of this dissertation research and avenues for future research are discussed.

5.1 Summary

The extant research has focused on supply risk and disruption management from a firm

perspective, however, a firm’s responsible decision makers play a key role in addressing

supply risk and disruptions. In the first chapter of this dissertation, three important but yet

unclear behavioral issues have been identified in this context. They have been addressed

by the studies presented in Chapters 2, 3, and 4. Their results are summarized in the

following.

5.1.1 Research question 1: Supply risk management

The increasing complexity and interconnectedness of modern supply chains have added

to the vulnerability of many firms to suffer from supply disruptions. Although most

supply chain managers are well-aware of their potentially severe negative consequences,

recent major disruptions have revealed that many supply chain managers prove

unprepared to effectively recover their firms’ operational performance (A.T. Kearney &

RapidRatings, 2018). Given that there is a need to better understand why some managers

proactively prepare for supply disruptions and others do not, Research Question 1 was

formulated.

To provide answers to Research Question 1, the research presented in Chapter 2

builds on PMT and its underlying idea of cognitive appraisal processes that determine an

individual’s intention to adopt a specific proactive measure. In addition, based on a

framework delineating key antecedents of proactive work behavior developed by Frese

and Fay (2001), this research accounts for the effects of certain individual-specific factors

on proactivity. By means of a carefully designed DCE, empirical data were collected to

explore the choice behavior of supply chain managers in the context of supply risk

management and evaluate the developed propositions. The data enabled an assessment of


the relative importance of certain proactive action attributes and a comparison of choice

behavior with the perceptions of the participants.

In line with the coping appraisal process of PMT, the DCE explored the relative

importance of (1) a specific proactive measure’s effectiveness (response efficacy), (2) an

individual’s ability to successfully implement the measure (self-efficacy), and (3) the

costs associated with pursuing the proactive measure (response costs). An analysis of the

results showed that all of these factors have significantly influenced the probability to

choose a specific proactive measure. Surprisingly, the effectiveness of a proactive

measure emerged as least important factor. When selecting proactive measures, response

costs (we distinguished between the following response costs: Direct implementation

costs and negative side effects on the respective buyer-supplier relationship) were

subconsciously given more importance than self-efficacy or response efficacy. However,

high vulnerability to a supply disruption mitigated the relative importance of response

costs.

A comparison of the subconsciously assigned weights with the consciously rated

perceived importance of these proactive action attributes unveiled a considerable

mismatch between decision makers’ perceptions and their choice behavior in the context

of supply risk management. While response efficacy was subconsciously assigned least

importance in the DCE, it was prioritized over response costs and self-efficacy when the

supply chain managers were asked to directly rate the relative importance of proactive

action attributes.

Further implications with regard to Research Question 1 were derived from the

influence of individual-specific factors on choice behavior. An individual’s proactive

personality, risk attitude, control appraisal, and experience had statistically significant

effects on the relative importance of the included proactive action attributes. These effects

shed light on what drives decision makers to or prevents them from selecting specific

proactive actions and the related insights were incorporated into an enriched model of

PMT that accounts for the influence of individual-specific factors on building protection

motivation (see Figure 5).

In sum, study 1 adds to a better understanding of why, in the supply risk context,

some managers take proactive action while others do not. This is an important

contribution as this is the first attempt to analyze choice behavior to derive insights into


proactive decision making in supply risk management. Moreover, several promising

opportunities for future research have been identified and discussed.

5.1.2 Research question 2: The influence of supply risk management on the impact

of disruptions

Although the topic of CSR has received considerable attention in business practice and

research, it is often merely associated with “doing good” while “avoiding bad” and, more

specifically, the issue of CSIR, have largely been neglected. In light of this, many firms

engage in CSR in the hopes of “insurance-like” effects in case of environmental or social

misconduct. However, the effects of ex ante CSR on negative stakeholder reactions

subsequent to CSIR are not fully understood. Thus, study 2 in Chapter 3 was concerned

with Research Question 2.

Two conflicting perspectives on the effect of a firm’s prior CSR activities on

negative stakeholder reactions to CSIR emanate from prior literature. On the one hand,

some studies have theorized an insurance-like mechanism such that prior CSR

engagement builds a “reservoir of goodwill” that mitigates negative responses to bad

news (e.g., Flammer, 2013; Godfrey et al., 2009). On the other hand, additional research

efforts conclude that firms which engage in CSR might create an expectation burden in

case of severe CSIR since these firms are often criticized the most (e.g., Swaen &

Vanhamme, 2003; Vanhamme & Grobben, 2009). To better understand whether these

effects depend on CSIR severity and the CSR reputation’s nature, study 2 presents and

discusses the results of a vignette-based experiment.

These results provide further empirical support for the circumstance that insurance-

like effects of CSR do not always hold and contribute to specifying their relevant

boundary conditions. Both CSIR severity and the CSR reputation’s nature (substantive/

symbolic) moderate the relationship between CSIR and negative stakeholder reactions.

Based on study 2, it can be concluded that substantive CSR reputations provide insurance-

like benefits in case of less severe CSIR but do not seem to mitigate negative stakeholder

reactions to severe CSIR. However, firms with reputations mainly driven by symbolic

CSR also experience insurance-like effects in response to less severe CSIR, but are

exposed to higher intentions to engage in nWOM after severe CSIR compared to firms

that did not promote themselves as socially responsible.


Being the first study that explicitly addressed the role of ex ante CSR reputations

on negative stakeholder reactions to CSIR in a B2B environment, the results contribute

to an improved understanding of CSIR-related risk and disruption management. In

addition, this study provides valuable empirical support for CSR engagement based on

instrumental rather than moral motives.

5.1.3 Research question 3: Supply disruption management

Supply disruptions pose considerable challenges to managers that seek to recover from

their negative consequences. After being exposed to supply disruptions, managers

typically face unreliable information about the consequences of possible response actions.

In this context, it is not trivial to decide on whether to wait until more reliable information

are available or directly launch response actions. Hence, to gain a better understanding of

how to effectively respond to supply disruptions, the research presented in Chapter 4 has

addressed Research Question 3 by means of a multi-method approach.

In a first step, simulation experiments with an agent-based model of supply

disruption recovery revealed how managers should behave in response to supply

disruptions. The model was developed based on complexity, response uncertainty, and

path dependence as main determinants of a supply chain manager’s environment in the

aftermath of a supply disruption. The results highlight that high complexity favors quick

action to be able to make sense of observable outcomes, but if response uncertainty is

high, more reliable information should be acquired before action is taken. Further key

insights were summarized in eight testable propositions which provide promising

opportunities for future research.

In a second step, the actual willingness of managers to take immediate action

subsequent to being affected by a supply disruption was explored. An analysis of data

from vignette-based experiments showed that response uncertainty and path dependence

reduced the participants’ intention to quickly respond to supply disruptions. Moreover,

although managers should actually respond quickly to supply disruptions in highly

complex environments if response uncertainty is not high, the results reveal that managers

tend to refrain from a quick response in case of high complexity. Hence, the multi-method

approach enabled the identification of a considerable mismatch between managers’

intentions and the recommended behavior based on the simulation experiments. Thereby,

Study 3 contributes to the emerging literature on supply disruption recovery and adds to


a better understanding of decision making behavior in the context of supply disruption

management.

5.2 Limitations

As with any empirical research, the results presented in this dissertation must be viewed

in light of certain overarching limitations concerning the data and the research designs.

First, the behavioral experiments conducted in Study 1 (discrete choice), Study 2

(vignette-based), and Study 3 (vignette-based) rely on intentions instead of actual

behavior as their outcome criteria. Stated intentions are considered the best predictors of

actual behavior (Ajzen, 1991; T. L. Webb & Sheeran, 2006). However, research has

shown that intentions are not perfectly correlated with future behavior, especially if there

is a large time lapse between measuring intentions and behavior or if an individual is

unable to act on an intention due to a lack of skills or unanticipated barriers (Fishbein,

2008; Morwitz, 1997). Hence, replicating the experiments with actual behavior as

outcome variable would enhance the explanatory power of the results presented.

Second, this dissertation focusses on individuals with centralized decision making

authority to investigate the behavior of purchasing professionals. Hence, the studies

conducted did not account for certain characteristics of industrial buying situations such

as hierarchical structures (W. J. Johnston & Bonoma, 1981), codes of conduct (Wotruba,

Chonko, & Loe, 2001), and the need to coordinate with other functions (Kocabasoglu &

Suresh, 2006). Thus, a worthwhile next step would be to develop research designs that

incorporate these characteristics and more comprehensively examine decision making

behavior in industrial buying situations.

Finally, the experiments with purchasing professionals conducted in Study 1, Study

2, and Study 3 mostly relied on respondents from German-speaking countries (Austria,

Germany, and Switzerland) which have relatively similar cultures (Hofstede, 1984,

2003). Their responses were treated as a single data set in the statistical analyses, since

previous research did not indicate any differences among these countries. However, it has

been shown that culture may influence behavior (e.g., Money et al., 1998). In addition,

the frequency and severity of natural disasters in Austria, Germany, and Switzerland are

very low compared to other countries or regions (e.g., Asia or the US) (Helferich &

Robert, 2002) which may have repercussions on the behavior of purchasing professionals.


Hence, replicating the experiments conducted as part of this dissertation research with

respondents from other cultural regions or from countries with different risk profiles

would be a promising next step to disentangle the influence of culture and environmental

conditions on choice behavior in the context of supply risks and supply disruptions.

5.3 Outlook

In addition to addressing the limitations mentioned above, several fruitful avenues for

future research in the area of supply risks and disruptions emerge from this dissertation

research.

First, the model of supply disruption recovery presented in Study 3 could serve as

a basis for further research on managing supply disruptions. Using an adjusted or

extended version of the model might provide additional insights into decision making

behavior of purchasing professionals. For example, although Study 3 focusses on whether

to delay a response subsequent to supply disruptions, the model could be used to

investigate the recovery performance implications of teams instead of single respondents

or account for different organizational structures. Moreover, the simulation experiments

in Study 3 have led to eight testable propositions which could be empirically investigated

by means of large-scale studies or dynamic experiments.

Second, although this dissertation provides important contributions to the academic

discussion on behavioral issues of managing supply risks and disruptions, many relevant

questions remain unanswered. Despite the growing scholarly interest in individual level

behavior in the supply risk literature, research on behavioral aspects of managing supply

risks and disruptions is still scant especially in light of the fact that coping with supply

risks poses several managerial challenges. Study 1 has highlighted that certain individual-

specific factors affect proactive risk management decisions. However, it is likely that

further personality-related factors, such as measures of ambition and extraversion, could

also considerably impact the decisions of the responsible managers. In addition, the

influence of an individual’s cultural background or organizational codes of conduct on

the behavior of purchasing professionals remain largely unclear. Given that many firms

have considerably increased their degree of globalization during the last decades and

comprise a multi-cultural set of employees, these issues might have important

repercussions on decision making in supply risk and disruption management.


Finally, this dissertation research strongly relies on experiments as methodological

approach to address the identified research questions. The studies’ results contribute

valuable insights into the behavior of purchasing professionals and, hence, highlight the

unique advantages of using experiments to study behavioral issues of supply risk and

disruption management such as control, efficiency, and responsiveness (Siemsen, 2011).

Nevertheless, supply chain researchers have only begun to exploit these benefits to, for

instance, augment or weaken the confidence in the validity of a theory by complementing

results of surveys or archival research with insights from behavioral experiments. Thus,

a promising avenue for future research on behavioral issues in supply risk and disruption

management is the intensified and innovative use of experiments.

References 101

References

A.T. Kearney, & RapidRatings. (2018). Managing supply risk: Are you prepared for a

black swan event? Retrieved from https://www.atkearney.com/analytics/article?

%2Fa%2Fmanaging-supply-risk-are-you-prepared-for-a-black-swan-event=&

utm_source=PRNewswire&utm_medium=PR&utm_campaign=2018americasat

kRR (accessed August 27, 2018).

Aguinis, H., & Bradley, K. J. (2014). Best practice recommendations for designing and

implementing experimental vignette methodology studies. Organizational

Research Methods, 17(4), 351-371.

Aguinis, H., Gottfredson, R. K., & Joo, H. (2013). Best-practice recommendations for

defining, identifying, and handling outliers. Organizational Research Methods,

16(2), 270-301.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human

Decision Processes, 50(2), 179-211.

Andersen, M., & Skjoett-Larsen, T. (2009). Corporate social responsibility in global

supply chains. Supply Chain Management: An International Journal, 14(2), 75-

86.

Anderson, E. J., Coltman, T., Devinney, T. M., & Keating, B. (2011). What drives the

choice of a third‐party logistics provider? Journal of Supply Chain Management,

47(2), 97-115.

Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994). Customer satisfaction, market

share, and profitability: Findings from Sweden. Journal of Marketing, 58(3), 53-

66.

Anderson, R. E. (1973). Consumer dissatisfaction: The effect of disconfirmed

expectancy on perceived product performance. Journal of Marketing Research,

10(1), 38-44.

Antonetti, P., & Maklan, S. (2016). An extended model of moral outrage at corporate

social irresponsibility. Journal of Business Ethics, 135(3), 429-444.

Armstrong, J. S. (1977). Social irresponsibility in management. Journal of Business

Research, 5(3), 185-213.

Arvidsson, S. (2010). Communication of corporate social responsibility: A study of the

views of management teams in large companies. Journal of Business Ethics,

96(3), 339-354.

Ashforth, B. E., & Gibbs, B. W. (1990). The double-edge of organizational legitimation.

Organization Science, 1(2), 177-194.

Aspinwall, L. G., & Taylor, S. E. (1997). A stitch in time: Self-regulation and proactive

coping. Psychological Bulletin, 121(3), 417-436.

Atzmüller, C., & Steiner, P. M. (2010). Experimental vignette studies in survey

research. Methodology, 6(3), 128-138.

https://www.atkearney.com/analytics/article?%2Fa%2Fmanaging-supply-risk-are-you-prepared-for-a-black-swan-event=&utm_source=PRNewswire&utm_medium=PR&utm_campaign=2018americasatkRR





References 102

Axelrod, L. J., & Newton, J. W. (1991). Preventing nuclear war: Beliefs and attitudes as

predictors of disarmist and deterrentist behavior. Journal of Applied Social

Psychology, 21(1), 29-40.

Bachrach, D. G., & Bendoly, E. (2011). Rigor in behavioral experiments: A basic

primer for supply chain management researchers. Journal of Supply Chain

Management, 47(3), 5-8.

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change.

Psychological Review, 84(2), 191-215.

Bandura, A. (2006). Toward a psychology of human agency. Perspectives on

Psychological Science, 1(2), 164-180.

Bansal, P., & Clelland, I. (2004). Talking trash: Legitimacy, impression management,

and unsystematic risk in the context of the natural environment. Academy of

Management Journal, 47(1), 93-103.

Bar-Eli, M., Azar, O. H., Ritov, I., Keidar-Levin, Y., & Schein, G. (2007). Action bias

among elite soccer goalkeepers: The case of penalty kicks. Journal of Economic

Psychology, 28(5), 606-621.

Barr, P. S. (1998). Adapting to unfamiliar environmental events: A look at the evolution

of interpretation and its role in strategic change. Organization Science, 9(6),

644-669.

Bateman, T. S., & Crant, J. M. (1993). The proactive component of organizational

behavior: A measure and correlates. Journal of Organizational Behavior, 14(2),

103-118.

Batson, C. D. (1998). Altruism and prosocial behavior. In S. T. Fiske, D. T. Gilbert, &

L. Gardner (Eds.), The Handbook of Social Psychology (pp. 282-316). New

York, NY: McGraw-Hill.

Beinhocker, E. D. (1999). Robust adaptive strategies. MIT Sloan Management Review,

40(3), 95-106.

Ben-Akiva, M. E., & Lerman, S. R. (1985). Discrete Choice Analysis: Theory and

Application to Travel Demand. Cambridge, MA: MIT Press.

Bénabou, R., & Tirole, J. (2010). Individual and corporate social responsibility.

Economica, 77(305), 1-19.

Bergkvist, L., & Rossiter, J. R. (2007). The predictive validity of multiple-item versus

single-item measures of the same constructs. Journal of Marketing Research,

44(2), 175-184.

Blackhurst, J., Craighead, C. W., Elkins, D., & Handfield, R. B. (2005). An empirically

derived agenda of critical research issues for managing supply-chain disruptions.

International Journal of Production Research, 43(19), 4067-4081.

Bode, C., & Macdonald, J. R. (2017). Stages of supply chain disruption response:

Direct, constraining, and mediating factors for impact mitigation. Decision

Sciences, 48(5), 836-874.

Bode, C., & Wagner, S. M. (2015). Structural drivers of upstream supply chain

complexity and the frequency of supply chain disruptions. Journal of Operations

Management, 36, 215-228.

References 103

Bode, C., Wagner, S. M., Petersen, K. J., & Ellram, L. M. (2011). Understanding

responses to supply chain disruptions: Insights from information processing and

resource dependence perspectives. Academy of Management Journal, 54(4),

833-856.

Boss, S. R., Galletta, D. F., Lowry, P. B., Moody, G. D., & Polak, P. (2015). What do

users have to fear? Using fear appeals to engender threats and fear that motivate

protective security behaviors. MIS Quarterly, 39(4), 837-864.

Boyle, K. J., Holmes, T. P., Teisl, M. F., & Roe, B. (2001). A comparison of conjoint

analysis response formats. American Journal of Agricultural Economics, 83(2),

441-454.

Bozic, B. (2017). Consumer trust repair: A critical literature review. European


Bradsher, K. (2017). Shivering children, pricier spandex: The impact of China’s energy

stumble. Retrieved from https://www.nytimes.com/2017/12/12/business/energy-

environment/china-gas-coal.html (accessed August 27, 2018).

Burger, L., & Sheahan, M. (2018). BASF shuts down U.S. TDI plant as feedstock

supply disrupted. Retrieved from https://www.reuters.com/article/basf-tdi/basf-

shuts-down-u-s-tdi-plant-as-feedstock-supply-disrupted-idUSFWN1PR0CN

(accessed August 27, 2018).

Burton, R., & Obel, B. (1984). Designing Efficient Organizations: Modelling and

Experimentation. Amsterdam, Netherlands: North-Holland.

Burton, R., & Obel, B. (1995). The validity of computational models in organization

science: From model realism to purpose of the model. Computational &

Mathematical Organization Theory, 1(1), 57-71.

Burton, R., & Obel, B. (2004). Strategic Organizational Diagnosis and Design: The

Dynamics of Fit. Boston, MA: Kluwer.

Business Continuity Institute. (2016). Supply Chain Resilience Report 2016.

Caversham, UK: Business Continuity Institute.

Campbell, J. L. (2007). Why would corporations behave in socially responsible ways?

An institutional theory of corporate social responsibility. Academy of

Management Review, 32(3), 946-967.

Cantor, D. E., Blackhurst, J. V., & Cortes, J. D. (2014). The clock is ticking: The role of

uncertainty, regulatory focus, and level of risk on supply chain disruption

decision making behavior. Transportation Research Part E: Logistics and

Transportation Review, 72, 159-172.

Carlos, W. C., & Lewis, B. W. (2018). Strategic silence: Withholding certification

status as a hypocrisy avoidance tactic. Administrative Science Quarterly, 63(1),

130-169.

Carroll, A. B. (1979). A three-dimensional conceptual model of corporate performance.

Academy of Management Review, 4(4), 497-505.

Chakravarthy, B. S. (1982). Adaptation: A promising metaphor for strategic

management. Academy of Management Review, 7(1), 35-44.

https://www.nytimes.com/2017/12/12/business/energy-environment/china-gas-coal.html

https://www.nytimes.com/2017/12/12/business/energy-environment/china-gas-coal.html

https://www.reuters.com/article/basf-tdi/basf-shuts-down-u-s-tdi-plant-as-feedstock-supply-disrupted-idUSFWN1PR0CN

https://www.reuters.com/article/basf-tdi/basf-shuts-down-u-s-tdi-plant-as-feedstock-supply-disrupted-idUSFWN1PR0CN

References 104

Chandrasekaran, A., Linderman, K., Sting, F. J., & Benner, M. J. (2016). Managing

R&D project shifts in high‐tech organizations: A multi‐method study.

Production and Operations Management, 25(3), 390-416.

Chen, M., Xia, Y., & Wang, X. (2010). Managing supply uncertainties through

Bayesian information update. IEEE Transactions on Automation Science and

Engineering, 7(1), 24-36.

Chen, Y., & Zahedi, F. M. (2016). Individuals' internet security perceptions and

behaviors: Polycontextual contrasts between the United States and China. MIS

Quarterly, 40(1), 205-222.

Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and

access to finance. Strategic Management Journal, 35(1), 1-23.

Choi, B., & La, S. (2013). The impact of corporate social responsibility (CSR) and

customer trust on the restoration of loyalty after service failure and recovery.

Journal of Services Marketing, 27(3), 223-233.

Christmann, P., & Taylor, G. (2006). Firm self-regulation through international

certifiable standards: Determinants of symbolic versus substantive

implementation. Journal of International Business Studies, 37(6), 863-878.

Christopher, M., & Peck, H. (2004). Building the resilient supply chain. The

International Journal of Logistics Management, 15(2), 1-14.

Cismaru, M., & Lavack, A. M. (2007). Interaction effects and combinatorial rules

governing protection motivation theory variables: A new model. Marketing

Theory, 7(3), 249-270.

Clark, A. (2010). Daimler settles US bribery charges. Retrieved from https://www.

theguardian.com/business/2010/mar/24/daimler-settles-us-bribery-charges


Colombo, S., Guerci, M., & Miandar, T. (2017). What do unions and employers

negotiate under the umbrella of Corporate Social Responsibility? Comparative

evidence from the Italian metal and chemical industries. Journal of Business

Ethics, 1-18.

Coltman, T., & Devinney, T. M. (2013). Modeling the operational capabilities for

customized and commoditized services. Journal of Operations Management,

31(7), 555-566.

Cooper, A. C., Folta, T. B., & Woo, C. (1995). Entrepreneurial information search.

Journal of Business Venturing, 10(2), 107-120.

Cox, G. (2000). Ready-Aim-Fire Problem Solving: A Strategic Approach to Decision

Making. Dublin, Ireland: Oak Tree Press.

Craig, C. S., & Douglas, S. P. (2005). International Marketing Research. Chichester,

UK: John Wiley & Sons.

Craighead, C. W., Blackhurst, J., Rungtusanatham, M. J., & Handfield, R. B. (2007).

The severity of supply chain disruptions: Design characteristics and mitigation

capabilities. Decision Sciences, 38(1), 131-156.

Cyert, R. M., & March, J. G. (1963). A Behavioral Theory of the Firm. Englewood

Cliffs, NJ: Prentice Hall.

https://www.theguardian.com/business/2010/mar/24/daimler-settles-us-bribery-charges



References 105

Daft, R. L., & Weick, K. E. (1984). Toward a model of organizations as interpretation

systems. Academy of Management Review, 9(2), 284-295.

de Jong, M. D., & van der Meer, M. (2017). How does it fit? Exploring the congruence

between organizations and their corporate social responsibility (CSR) activities.

Journal of Business Ethics, 143(1), 71-83.

Demuijnck, G., & Fasterling, B. (2016). The social license to operate. Journal of

Business Ethics, 136(4), 675-685.

Dhar, R. (1997). Consumer preference for a no-choice option. Journal of Consumer

Research, 24(2), 215-231.

Dill, W. R. (1958). Environment as an influence on managerial autonomy.

Administrative Science Quarterly, 2(4), 409-443.

Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J., & Wagner, G. G. (2011).

Individual risk attitudes: Measurement, determinants, and behavioral

consequences. Journal of the European Economic Association, 9(3), 522-550.

Donia, M. B., Ronen, S., Sirsly, C.-A. T., & Bonaccio, S. (2017). CSR by any other

name? The differential impact of substantive and symbolic CSR attributions on

employee outcomes. Journal of Business Ethics, 1-21.

Dubrovski, D. (2004). Peculiarities of managing a company in crisis. Total Quality

Management & Business Excellence, 15(9/10), 1199-1207.

Duncan, R. B. (1972). Characteristics of organizational environments and perceived

environmental uncertainty. Administrative Science Quarterly, 17(3), 313-327.

Durand, R., Hawn, O., & Ioannou, I. (2017). Willing and able: A general model of

organizational responses to normative pressures. Academy of Management

Review, forthcoming.

Dutton, J. E., Fahey, L., & Narayanan, V. K. (1983). Toward understanding strategic

issue diagnosis. Strategic Management Journal, 4(4), 307-323.

Eccles, R. G., Ioannou, I., & Serafeim, G. (2014). The impact of corporate sustainability

on organizational processes and performance. Management Science, 60(11),

2835-2857.

Einwiller, S. A., Fedorikhin, A., Johnson, A. R., & Kamins, M. A. (2006). Enough is

enough! When identification no longer prevents negative corporate associations.

Journal of the Academy of Marketing Science, 34(2), 185-194.

Ethiraj, S. K., & Levinthal, D. (2004). Bounded rationality and the search for

organizational architecture: An evolutionary perspective on the design of

organizations and their evolvability. Administrative Science Quarterly, 49(3),

404-437.

Fay, D., & Frese, M. (2001). The concept of personal initiative: An overview of validity

studies. Human Performance, 14(1), 97-124.

Ferguson, J. L., & Johnston, W. J. (2011). Customer response to dissatisfaction: A

synthesis of literature and conceptual framework. Industrial Marketing

Management, 40(1), 118-127.

References 106

Fiaschi, D., Giuliani, E., & Nieri, F. (2017). Overcoming the liability of origin by doing

no-harm: Emerging country firms’ social irresponsibility as they go global.

Journal of World Business, 52(4), 546-563.

Fishbein, M. (2008). A reasoned action approach to health promotion. Medical Decision

Making, 28(6), 834-844.

Fisher, R. J. (1993). Social desirability bias and the validity of indirect questioning.

Journal of Consumer Research, 20(2), 303-315.

Flammer, C. (2013). Corporate social responsibility and shareholder reaction: The

environmental awareness of investors. Academy of Management Journal, 56(3),

758-781.

Floyd, D. L., Prentice‐Dunn, S., & Rogers, R. W. (2000). A meta‐analysis of research

on protection motivation theory. Journal of Applied Social Psychology, 30(2),

407-429.

Forehand, M. R., & Grier, S. (2003). When is honesty the best policy? The effect of

stated company intent on consumer skepticism. Journal of Consumer

Psychology, 13(3), 349-356.

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with

unobservable variables and measurement error. Journal of Marketing Research,

18(1), 39-50.

Freeman, I., & Hasnaoui, A. (2011). The meaning of corporate social responsibility:

The vision of four nations. Journal of Business Ethics, 100(3), 419-443.

Frese, M., & Fay, D. (2001). 4. Personal initiative: An active performance concept for

work in the 21st century. Research in Organizational Behavior, 23, 133-187.

Galbraith, J. (1973). Designing Complex Organizations. Reading, MA: Addison-

Wesley.

Galbraith, J. (1977). Organization Design. Reading, MA: Addison-Wesley.

Gavetti, G., Levinthal, D. A., & Rivkin, J. W. (2005). Strategy making in novel and

complex worlds: The power of analogy. Strategic Management Journal, 26(8),

691-712.

Germann, F., Grewal, R., Ross, W. T., Jr, & Srivastava, R. K. (2014). Product recalls

and the moderating role of brand commitment. Marketing Letters, 25(2), 179-

191.

Godfrey, P. C., Merrill, C. B., & Hansen, J. M. (2009). The relationship between

corporate social responsibility and shareholder value: An empirical test of the

risk management hypothesis. Strategic Management Journal, 30(4), 425-445.

Grant, A. M., & Ashford, S. J. (2008). The dynamics of proactivity at work. Research in

Organizational Behavior, 28, 3-34.

Grant, A. M., Parker, S., & Collins, C. (2009). Getting credit for proactive behavior:

Supervisor reactions depend on what you value and how you feel. Personnel

Psychology, 62(1), 31-55.

Green, C., & Gerard, K. (2009). Exploring the social value of health‐care interventions:

A stated preference discrete choice experiment. Health Economics, 18(8), 951-

976.

References 107

Greenpeace. (2014). A monstrous mess: Toxic water pollution in China. Retrieved from

http://www.greenpeace.org/international/en/news/features/A-Monstrous-Mess-

toxic-water-pollution-in-China/ (accessed August 27, 2018).

Griffin, M. A., Neal, A., & Parker, S. K. (2007). A new model of work role

performance: Positive behavior in uncertain and interdependent contexts.

Academy of Management Journal, 50(2), 327-347.

Grothmann, T., & Reusswig, F. (2006). People at risk of flooding: Why some residents

take precautionary action while others do not. Natural Hazards, 38(1), 101-120.

Gualandris, J., Klassen, R. D., Vachon, S., & Kalchschmidt, M. (2015). Sustainable

evaluation and verification in supply chains: Aligning and leveraging

accountability to stakeholders. Journal of Operations Management, 38, 1-13.

Habermann, M., Blackhurst, J., & Metcalf, A. Y. (2015). Keep your friends close?

Supply chain design and disruption risk. Decision Sciences, 46(3), 491-526.

Hada, M., Grewal, R., & Lilien, G. L. (2014). Supplier-selected referrals. Journal of

Marketing, 78(2), 34-51.

Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2009).

Multivariate Data Analysis (7th ed.). Upper Saddle River, NJ: Prentice Hall.

Hajmohammad, S., & Vachon, S. (2016). Mitigation, avoidance, or acceptance?

Managing supplier sustainability risk. Journal of Supply Chain Management,

52(2), 48-65.

Hale, J. E., Hale, D. P., & Dulek, R. E. (2006). Decision processes during crisis

response: An exploratory investigation. Journal of Managerial Issues, 18(3),

301-320.

Ham, C.-D., & Kim, J. (2017). The role of CSR in crises: Integration of situational

crisis communication theory and the persuasion knowledge model. Journal of

Business Ethics, 1-20.

Handfield, R. B., Blackhurst, J., Elkins, D., & Craighead, C. W. (2007). A framework

for reducing the impact of disruptions to the supply chain: Observations from

multiple executives. In R. B. Handfield & K. McCormack (Eds.), Supply Chain

Risk Management (pp. 29-49). New York, NY: Taylor and Francis.

Hartmann, J., & Moeller, S. (2014). Chain liability in multitier supply chains?

Responsibility attributions for unsustainable supplier behavior. Journal of

Operations Management, 32(5), 281-294.

Hawn, O., & Ioannou, I. (2012). Do actions speak louder than words? The case of

corporate social responsibility (CSR). Academy of Management Proceedings

2012.

Hawn, O., & Ioannou, I. (2016). Mind the gap: The interplay between external and

internal actions in the case of corporate social responsibility. Strategic


Heckmann, I., Comes, T., & Nickel, S. (2015). A critical review on supply chain risk –

Definition, measure and modeling. Omega, 52, 119-132.

Heide, J. B., Wathne, K. H., & Rokkan, A. I. (2007). Interfirm monitoring, social

contracts, and relationship outcomes. Journal of Marketing Research, 44(3),

425-433.

http://www.greenpeace.org/international/en/news/features/A-Monstrous-Mess-toxic-water-pollution-in-China/

http://www.greenpeace.org/international/en/news/features/A-Monstrous-Mess-toxic-water-pollution-in-China/

References 108

Helferich, O. K., & Robert, L. C. (2002). Securing the Supply Chain. Oak Brook, IL:

Council of Logistics Management.

Helft, M., & Bunkley, N. (2011). Disaster in Japan batters suppliers. Retrieved from

http://www.nytimes.com/2011/03/15/business/global/15supply.html?mcubz=1


Hendricks, K. B., & Singhal, V. R. (2005a). Association between supply chain glitches

and operating performance. Management Science, 51(5), 695-711.

Hendricks, K. B., & Singhal, V. R. (2005b). An empirical analysis of the effect of

supply chain disruptions on long-run stock price performance and equity risk of

the firm. Production and Operations Management, 14(1), 35-52.

Hensher, D. A., Rose, J. M., & Greene, W. H. (2005). Applied Choice Analysis: A

Primer. New York, NY: Cambridge University Press.

Hepher, T., & Altmeyer, C. (2017). Airbus first-quarter profit slides on engine delays,

price squeeze. Retrieved from https://www.reuters.com/article/us-airbus-results-

idUSKBN17T0JE (accessed August 27, 2018).

Highhouse, S., Brooks, M. E., & Gregarus, G. (2009). An organizational impression

management perspective on the formation of corporate reputations. Journal of

Management, 35(6), 1481-1493.

Hofmann, H., Busse, C., Bode, C., & Henke, M. (2014). Sustainability‐related supply

chain risks: Conceptualization and management. Business Strategy and the

Environment, 23(3), 160-172.

Hofstede, G. (1984). Culture's Consequences: International Differences in Work-

Related Values. Beverly Hills, CA: Sage Publications.

Hofstede, G. (1993). Cultural constraints in management theories. The Executive, 7(1),

81-94.

Hofstede, G. (2003). Culture's Consequences: Comparing Values, Behaviors,

Institutions, and Organizations Across Nations. Thousand Oaks, CA: Sage

Publications.

Homburg, C., Stierl, M., & Bornemann, T. (2013). Corporate social responsibility in

business-to-business markets: How organizational customers account for

supplier corporate social responsibility engagement. Journal of Marketing,

77(6), 54-72.

Hora, M., & Klassen, R. D. (2013). Learning from others’ misfortune: Factors

influencing knowledge acquisition to reduce operational risk. Journal of

Operations Management, 31(1), 52-61.

Hovland, C. I., Harvey, O., & Sherif, M. (1957). Assimilation and contrast effects in

reactions to communication and attitude change. Journal of Abnormal and

Social Psychology, 55(2), 244-252.

Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and Persuasion.

New Haven, CT: Yale University Press.

Hunter, J. E. (1986). Cognitive ability, cognitive aptitudes, job knowledge, and job

performance. Journal of Vocational Behavior, 29(3), 340-362.

http://www.nytimes.com/2011/03/15/business/global/15supply.html?mcubz=1

https://www.reuters.com/article/us-airbus-results-idUSKBN17T0JE

https://www.reuters.com/article/us-airbus-results-idUSKBN17T0JE

References 109

Huq, F. A., Chowdhury, I. N., & Klassen, R. D. (2016). Social management capabilities

of multinational buying firms and their emerging market suppliers: An

exploratory study of the clothing industry. Journal of Operations Management,

46, 19-37.

Isenberg, D. J. (1984). How senior managers think. Harvard Business Review, 84(6),

81-90.

Isenberg, D. J. (1986). Thinking and managing: A verbal protocol analysis of

managerial problem solving. Academy of Management Journal, 29(4), 775-788.

Janssen, C., Sen, S., & Bhattacharya, C. (2015). Corporate crises in the age of corporate

social responsibility. Business Horizons, 58(2), 183-192.

Jin, H. (2016). VW hit with £12m fine and sales ban on 80 models in South Korea.

Retrieved from http://www.independent.co.uk/news/business/news/vw-

volkswagen-south-korea-sales-ban-emissions-scandal-fine-audi-bentley-

dieselgate-a7167601.html (accessed August 27, 2018).

Johnston, A. C., & Warkentin, M. (2010). Fear appeals and information security

behaviors: An empirical study. MIS Quarterly, 34(3), 549-566.

Johnston, W. J., & Bonoma, T. V. (1981). The buying center: Structure and interaction

patterns. Journal of Marketing, 45(3), 143-156.

Jones, G. H., Jones, B. H., & Little, P. (2000). Reputation as reservoir: Buffering

against loss in times of economic crisis. Corporate Reputation Review, 3(1), 21-

29.

Kang, C., Germann, F., & Grewal, R. (2016). Washing away your sins? Corporate

social responsibility, corporate social irresponsibility, and firm performance.

Journal of Marketing, 80(2), 59-79.

Katila, R., & Ahuja, G. (2002). Something old, something new: A longitudinal study of

search behavior and new product introduction. Academy of Management

Journal, 45(6), 1183-1194.

Kauffman, S. A. (1993). The Origins of Order: Self-Organization and Selection in

Evolution. New York, NY: Oxford University Press.

Kaul, A., & Luo, J. (2018). An economic case for CSR: The comparative efficiency of

for‐profit firms in meeting consumer demand for social goods. Strategic


Kepner, C. H., & Tregoe, B. B. (1965). The Rational Manager. New York, NY:

McGraw-Hill.

Kerstholt, J. H. (1996). The effect of information costs on strategy selection in dynamic

tasks. Acta Psychologica, 94(3), 273-290.

Ketola, T. (2010). Five leaps to corporate sustainability through a corporate

responsibility portfolio matrix. Corporate Social Responsibility and

Environmental Management, 17(6), 320-336.

Kim, S., & Choi, S. M. (2016). Congruence effects in post-crisis CSR communication:

The mediating role of attribution of corporate motives. Journal of Business

Ethics, 1-17.

http://www.independent.co.uk/news/business/news/vw-volkswagen-south-korea-sales-ban-emissions-scandal-fine-audi-bentley-dieselgate-a7167601.html



References 110

Kim, Y. H., & Davis, G. F. (2016). Challenges for global supply chain sustainability:

Evidence from conflict minerals reports. Academy of Management Journal,

59(6), 1896-1916.

Klein, J., & Dawar, N. (2004). Corporate social responsibility and consumers'

attributions and brand evaluations in a product–harm crisis. International

Journal of Research in Marketing, 21(3), 203-217.

Kleinmuntz, D. N., & Thomas, J. B. (1987). The value of action and inference in

dynamic decision making. Organizational Behavior and Human Decision

Processes, 39(3), 341-364.

Klir, G. J. (2005). Uncertainty and Information: Foundations of Generalized

Information Theory. Hoboken, NJ: John Wiley & Sons.

Knemeyer, A. M., Zinn, W., & Eroglu, C. (2009). Proactive planning for catastrophic

events in supply chains. Journal of Operations Management, 27(2), 141-153.

Knudsen, T., & Levinthal, D. A. (2007). Two faces of search: Alternative generation

and alternative evaluation. Organization Science, 18(1), 39-54.

Kocabasoglu, C., & Suresh, N. C. (2006). Strategic sourcing: An empirical investigation

of the concept and its practices in US manufacturing firms. Journal of Supply

Chain Management, 42(2), 4-16.

Krause, D. R., Vachon, S., & Klassen, R. D. (2009). Special topic forum on sustainable

supply chain management: Introduction and reflections on the role of purchasing

management. Journal of Supply Chain Management, 45(4), 18-25.

Lancsar, E., & Louviere, J. J. (2008). Conducting discrete choice experiments to inform

healthcare decision making. Pharmacoeconomics, 26(8), 661-677.

Latour, A. (2001). Trial by fire: A blaze in Albuquerque sets off major crisis for cell-

phone giants. Wall Street Journal, January 29, 2001.

Lee, H. L. (2004). The triple-A supply chain. Harvard Business Review, 82(10), 102-

113.

Lenz, I., Wetzel, H. A., & Hammerschmidt, M. (2017). Can doing good lead to doing

poorly? Firm value implications of CSR in the face of CSI. Journal of the

Academy of Marketing Science, 45(5), 677-697.

Leroi-Werelds, S., Streukens, S., Brady, M. K., & Swinnen, G. (2014). Assessing the

value of commonly used methods for measuring customer value: A multi-setting

empirical study. Journal of the Academy of Marketing Science, 42(4), 430-451.

Levinthal, D. A. (1997). Adaptation on rugged landscapes. Management Science, 43(7),

934-950.

Liebowitz, S., & Margolis, S. (2000). Path dependence. In B. Bouckaert & G. De Geest

(Eds.), Encyclopedia of Law and Economics (pp. 981-998). Cheltenham, UK;

Northampton, MA: Edward Elgar.

Lin-Hi, N., & Blumberg, I. (2018). The link between (not) practicing CSR and

corporate reputation: Psychological foundations and managerial implications.


Lin-Hi, N., & Müller, K. (2013). The CSR bottom line: Preventing corporate social

irresponsibility. Journal of Business Research, 66(10), 1928-1936.

References 111

Lipshitz, R., & Strauss, O. (1997). Coping with uncertainty: A naturalistic decision-

making analysis. Organizational Behavior and Human Decision Processes,

69(2), 149-163.

Lounamaa, P. H., & March, J. G. (1987). Adaptive coordination of a learning team.

Management Science, 33(1), 107-123.

Louviere, J. J., Flynn, T. N., & Carson, R. T. (2010). Discrete choice experiments are

not conjoint analysis. Journal of Choice Modelling, 3(3), 57-72.

Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000). Stated Choice Methods: Analysis

and Applications. Cambridge, UK: Cambridge University Press.

Louviere, J. J., & Woodworth, G. (1983). Design and analysis of simulated consumer

choice or allocation experiments: An approach based on aggregate data. Journal

of Marketing Research, 20(4), 350-367.

Lynes, J. K., & Andrachuk, M. (2008). Motivations for corporate social and

environmental responsibility: A case study of Scandinavian Airlines. Journal of

International Management, 14(4), 377-390.

Lynn, M. (2015). Corporate Social Responsibility has become a racket - and a

dangerous one. Retrieved from http://www.telegraph.co.uk/finance/

newsbysector/industry/11896546/Corporate-Social-Responsibility-has-become-

a-racket-and-a-dangerous-one.html (accessed August 27, 2018).

Lyon, T. P., & Maxwell, J. W. (2011). Greenwash: Corporate environmental disclosure

under threat of audit. Journal of Economics and Management Strategy, 20(1), 3-

41.

Macdonald, J. R., & Corsi, T. M. (2013). Supply chain disruption management: Severe

events, recovery, and performance. Journal of Business Logistics, 34(4), 270-

288.

Mantel, S. P., Tatikonda, M. V., & Liao, Y. (2006). A behavioral study of supply

manager decision-making: Factors influencing make versus buy evaluation.

Journal of Operations Management, 24(6), 822-838.

March, J. G., & Shapira, Z. (1987). Managerial perspectives on risk and risk taking.

Management Science, 33(11), 1404-1418.

March, J. G., & Simon, H. A. (1958). Organizations. New York, NY: John Wiley &

Sons.

Matejek, S., & Gössling, T. (2014). Beyond legitimacy: A case study in BP’s “green

lashing”. Journal of Business Ethics, 120(4), 571-584.

Maxham, J. G., III, & Netemeyer, R. G. (2003). Firms reap what they sow: The effects

of shared values and perceived organizational justice on customers’ evaluations

of complaint handling. Journal of Marketing, 67(1), 46-62.

McFadden, D. (1986). The choice theory approach to market research. Marketing

Science, 5(4), 275-297.

McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & E. Aronson

(Eds.), Handbook of Social Psychology (pp. 233-346). New York, NY: Random

House.

http://www.telegraph.co.uk/finance/newsbysector/industry/11896546/Corporate-Social-Responsibility-has-become-a-racket-and-a-dangerous-one.html




References 112

McKelvie, A., Haynie, J. M., & Gustavsson, V. (2011). Unpacking the uncertainty

construct: Implications for entrepreneurial action. Journal of Business

Venturing, 26(3), 273-292.

McMullen, J. S., & Shepherd, D. A. (2006). Entrepreneurial action and the role of

uncertainty in the theory of the entrepreneur. Academy of Management Review,

31(1), 132-152.

McShane, L., & Cunningham, P. (2012). To thine own self be true? Employees’

judgments of the authenticity of their organization’s corporate social

responsibility program. Journal of Business Ethics, 108(1), 81-100.

McWilliams, A., & Siegel, D. (2001). Corporate social responsibility: A theory of the

firm perspective. Academy of Management Review, 26(1), 117-127.

Mehrotra, K. (2018). BMW sued by U.S. diesel drivers over emissions-test cheating.

Retrieved from https://www.bloomberg.com/news/articles/2018-03-27/bmw-

sued-for-installing-defeat-devices-in-u-s-diesel-cars (accessed August 27, 2018).

Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as

myth and ceremony. American Journal of Sociology, 83(2), 340-363.

Mihm, J., Loch, C. H., Wilkinson, D., & Huberman, B. A. (2010). Hierarchical structure

and search in complex organizations. Management Science, 56(5), 831-848.

Milliken, F. J. (1987). Three types of perceived uncertainty about the environment:

State, effect, and response uncertainty. Academy of Management Review, 12(1),

133-143.

Minor, D., & Morgan, J. (2011). CSR as reputation insurance: Primum non nocere.

California Management Review, 53(3), 40-59.

Mintzberg, H. (1973). The Nature of Managerial Work. New York, NY: Harper & Row

Publishers.

Mintzberg, H., Raisinghani, D., & Theoret, A. (1976). The structure of "unstructured"

decision processes. Administrative Science Quarterly, 21(2), 246-275.

Mintzberg, H., & Westley, F. (2001). Decision making: It's not what you think. MIT

Sloan Management Review, 42(3), 89-93.

Mitroff, I. I., & Alpaslan, M. C. (2003). Preparing for evil. Harvard Business Review,

81(4), 109-115.

Money, R. B., Gilly, M. C., & Graham, J. L. (1998). Explorations of national culture

and word-of-mouth referral behavior in the purchase of industrial services in the

United States and Japan. Journal of Marketing, 62(4), 76-87.

Moore, W. L., Gray-Lee, J., & Louviere, J. J. (1998). A cross-validity comparison of

conjoint analysis and choice models at different levels of aggregation. Marketing

Letters, 9(2), 195-207.

Morgan, N. A., & Rego, L. L. (2006). The value of different customer satisfaction and

loyalty metrics in predicting business performance. Marketing Science, 25(5),

426-439.

Moriyasu, K. (2017). BMW closes major China assembly line for the week. Retrieved

from http://asia.nikkei.com/Business/Companies/BMW-closes-major-China-

assembly-line-for-the-week (accessed August 27, 2018).

https://www.bloomberg.com/news/articles/2018-03-27/bmw-sued-for-installing-defeat-devices-in-u-s-diesel-cars

https://www.bloomberg.com/news/articles/2018-03-27/bmw-sued-for-installing-defeat-devices-in-u-s-diesel-cars

http://asia.nikkei.com/Business/Companies/BMW-closes-major-China-assembly-line-for-the-week

http://asia.nikkei.com/Business/Companies/BMW-closes-major-China-assembly-line-for-the-week

References 113

Morwitz, V. (1997). Why consumers don't always accurately predict their own future

behavior. Marketing Letters, 8(1), 57-70.

Munich Re. (2015). NAT CATS 2014: What's going on with the weather? Retrieved

from https://www.munichre.com/site/mram/get/documents_E-1959049670/

mram/assetpool.munichreamerica.wrap/PDF/2014/MunichRe_III_NatCatWebin

ar_01072015w.pdf (accessed August 27, 2018).

Munich Re. (2017). Natural catastrophes 2016. Retrieved from https://www.munichre.c

om/site/mram-mobile/get/documents_E-253943638/mram/assetpool.mr_

america/PDFs/3_Publications/TOPICS_GEO/TOPICS_GEO_2017.pdf


Nelson, D. R., Adger, W. N., & Brown, K. (2007). Adaptation to environmental change:

Contributions of a resilience framework. Annual Review of Environment and

Resources, 32(1), 395-419.

Noda, T., & Collis, D. J. (2001). The evolution of intraindustry firm heterogeneity:

Insights from a process study. Academy of Management Journal, 44(4), 897-

925.

Norman, G. (2010). Likert scales, levels of measurement and the “laws” of statistics.

Advances in Health Sciences Education, 15(5), 625-632.

Norrman, A., & Jansson, U. (2004). Ericsson's proactive supply chain risk management

approach after a serious sub-supplier accident. International Journal of Physical

Distribution and Logistics Management, 34(5), 434-456.

O'Reilly, C. A. (1982). Variations in decision makers' use of information sources: The

impact of quality and accessibility of information. Academy of Management

Journal, 25(4), 756-771.

Oliver, C. (1991). Strategic responses to institutional processes. Academy of


Parker, J. R., & Schrift, R. Y. (2011). Rejectable choice sets: How seemingly irrelevant

no-choice options affect consumer decision processes. Journal of Marketing

Research, 48(5), 840-854.

Parker, S. K., Williams, H. M., & Turner, N. (2006). Modeling the antecedents of

proactive behavior at work. Journal of Applied Psychology, 91(3), 636-652.

Paulraj, A., Chen, I. J., & Blome, C. (2017). Motives and performance outcomes of

sustainable supply chain management practices: A multi-theoretical perspective.


Pereira, C. R., Christopher, M., & Lago Da Silva, A. (2014). Achieving supply chain

resilience: The role of procurement. Supply Chain Management: An

International Journal, 19(5/6), 626-642.

Peters, T. J., & Waterman, R. H. (1982). In Search of Excellence: Lessons from

America's best-run Companies. New York, NY: Harper & Row.

Potoglou, D., & Kanaroglou, P. S. (2007). Household demand and willingness to pay

for clean vehicles. Transportation Research Part D: Transport and

Environment, 12(4), 264-274.

https://www.munichre.com/site/mram/get/documents_E-1959049670/mram/assetpool.munichreamerica.wrap/PDF/2014/MunichRe_III_NatCatWebinar_01072015w.pdf




https://www.munichre.com/site/mram-mobile/get/documents_E-253943638/mram/assetpool.mr_america/PDFs/3_Publications/TOPICS_GEO/TOPICS_GEO_2017.pdf




References 114

Pullman, M. E., Verma, R., & Goodale, J. C. (2001). Service design and operations

strategy formulation in multicultural markets. Journal of Operations

Management, 19(2), 239-254.

Raaijmakers, A. G., Vermeulen, P. A., Meeus, M. T., & Zietsma, C. (2015). I need

time! Exploring pathways to compliance under institutional complexity.

Academy of Management Journal, 58(1), 85-110.

Ramasamy, B., Yeung, M. C., & Au, A. K. (2010). Consumer support for corporate

social responsibility (CSR): The role of religion and values. Journal of Business

Ethics, 91(1), 61-72.

Rao, S., & Goldsby, T. J. (2009). Supply chain risks: A review and typology.

International Journal of Logistics Management, 20(1), 97-123.

Reuters. (2014). Wal-Mart must pay $188 million in workers' class action. Retrieved

from http://www.reuters.com/article/us-walmart-lawsuit-idUSKBN0JU1X

J20141216 (accessed August 27, 2018).

Richins, M. L. (1983). Negative word-of-mouth by dissatisfied consumers: A pilot

study. Journal of Marketing, 47(1), 68-78.

Ro, Y. K., Su, H. C., & Chen, Y. S. (2016). A tale of two perspectives on an impending

supply disruption. Journal of Supply Chain Management, 52(1), 3-20.

Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude

change. The Journal of Psychology, 91(1), 93-114.

Rogers, R. W. (1983). Cognitive and psychological processes in fear appeals and

attitude change: A revised theory of protection motivation. In J. Cacioppo & R.

Petty (Eds.), Social Psychophysiology (pp. 153-176). New York, NY: Guildford

Press.

Rogers, R. W., & Prentice-Dunn, S. (1997). Protection motivation theory. In D. S.

Gochman (Ed.), Handbook of Health Behavior Research I: Personal and Social

Determinants (pp. 113-132). New York, NY: Plenum Press.

Rudolph, J. W., Morrison, J. B., & Carroll, J. S. (2009). The dynamics of action-

oriented problem solving: Linking interpretation and choice. Academy of


Rungtusanatham, M., Wallin, C., & Eckerd, S. (2011). The vignette in a scenario‐based

role‐playing experiment. Journal of Supply Chain Management, 47(3), 9-16.

Russo, M. V., & Harrison, N. S. (2005). Organizational design and environmental

performance: Clues from the electronics industry. Academy of Management

Journal, 48(4), 582-593.

Ryan, M., Gerard, K., & Amaya-Amaya, M. (2007). Using Discrete Choice

Experiments to Value Health and Health Care. Dordrecht, Netherlands:

Springer.

Ryan, M., & Skåtun, D. (2004). Modelling non‐demanders in choice experiments.

Health Economics, 13(4), 397-402.

Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal

of Risk and Uncertainty, 1(1), 7-59.

http://www.reuters.com/article/us-walmart-lawsuit-idUSKBN0JU1XJ20141216

http://www.reuters.com/article/us-walmart-lawsuit-idUSKBN0JU1XJ20141216

References 115

Scherer, K. R. (1984). Emotion as a multicomponent process: A model and some cross-

cultural data. In P. Scherer (Ed.), Review of Personality and Social Psychology

(pp. 37-63). Beverly Hills, CA: Sage Publications.

Scherer, K. R. (1988). Criteria for emotion-antecedent appraisal: A review. In V.

Hamiltion, G. H. Bower, & N. H. Frijda (Eds.), Cognitive Perspectives on

Emotion and Motivation (pp. 89-126). Norwell, MA: Kluwer Academic

Publishers.

Schmidt, C. G., Foerstl, K., & Schaltenbrand, B. (2017). The supply chain position

paradox: Green practices and firm performance. Journal of Supply Chain

Management, 53(1), 3-25.

Schmidt, F. L., Hunter, J. E., & Outerbridge, A. N. (1986). Impact of job experience and

ability on job knowledge, work sample performance, and supervisory ratings of

job performance. Journal of Applied Psychology, 71(3), 432-439.

Schneider, L., & Wallenburg, C. M. (2012). Implementing sustainable sourcing—Does

purchasing need to change? Journal of Purchasing and Supply Management,

18(4), 243-257.

Schneider, S. C., & De Meyer, A. (1991). Interpreting and responding to strategic

issues: The impact of national culture. Strategic Management Journal, 12(4),

307-320.

Seibert, S. E., Crant, J. M., & Kraimer, M. L. (1999). Proactive personality and career

success. Journal of Applied Psychology, 84(3), 416-427.

Sen, S., & Bhattacharya, C. B. (2001). Does doing good always lead to doing better?

Consumer reactions to corporate social responsibility. Journal of Marketing

Research, 38(2), 225-243.

Sheehy, B. (2015). Defining CSR: Problems and solutions. Journal of Business Ethics,

131(3), 625-648.

Sheeran, P., Harris, P. R., & Epton, T. (2013). Does heightening risk appraisals change

people’s intentions and behavior? A meta-analysis of experimental studies.

Psychological Bulletin, 140(2), 511-543.

Sheffi, Y. (2005). The Resilient Enterprise: Overcoming Vulnerability for Competitive

Advantage. Cambridge, MA: MIT Press.

Sherif, M., & Hovland, C. I. (1961). Social Judgment: Assimilation and Contrast Effects

in Communication and Attitude Change. New Haven, CT: Yale University

Press.

Shiu, Y. M., & Yang, S. L. (2017). Does engagement in corporate social responsibility

provide strategic insurance‐like effects? Strategic Management Journal, 38(2),

455-470.

Siemsen, E. (2011). The usefulness of behavioral laboratory experiments in supply

chain management research. Journal of Supply Chain Management, 47(3), 17-

18.

Siggelkow, N., & Rivkin, J. W. (2005). Speed and search: Designing organizations for

turbulence and complexity. Organization Science, 16(2), 101-122.

Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of

Economics, 69(1), 99-118.

References 116

Simon, H. A. (1962). The architecture of complexity. Proceedings of the American

Philosophical Society, 106(6), 467-482.

Simon, H. A. (1979). Rational decision making in business organizations. The American

Economic Review, 69(4), 493-513.

Sims, R. R., & Brinkmann, J. (2003). Enron ethics (or: Culture matters more than

codes). Journal of Business Ethics, 45(3), 243-256.

Singh, J. (1990). Voice, exit, and negative word-of-mouth behaviors: An investigation

across three service categories. Journal of the Academy of Marketing Science,

18(1), 1-15.

Skellett, B., Cairns, B., Geard, N., Tonkes, B., & Wiles, J. (2005). Maximally rugged

NK landscapes contain the highest peaks. Paper presented at the Proceedings of

the 2005 Genetic and Evolutionary Computation Conference.

Sodhi, M. S., Son, B. G., & Tang, C. S. (2012). Researchers' perspectives on supply

chain risk management. Production and Operations Management, 21(1), 1-13.

Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches.

Academy of Management Review, 20(3), 571-610.

Swaen, V., & Vanhamme, J. (2003). Do accusations of irresponsible acts hurts

companies more when they promote themselves as socially responsible? In G.

D. Laurent, D. Merunka, & J. Zaichowsky (Eds.), Marketing Communications

and Consumer Behavior Proceedings (pp. 175-188).

Sydow, J., Schreyögg, G., & Koch, J. (2009). Organizational path dependence: Opening

the black box. Academy of Management Review, 34(4), 689-709.

Tajitsu, N., & Yamazaki, M. (2016). Toyota, other major Japanese firms hit by quake

damage, supply disruptions. Retrieved from https://www.reuters.com/article/us-

japan-quake-toyota/toyota-other-major-japanese-firms-hit-by-quake-damage-

supply-disruptions-idUSKCN0XE08O (accessed August 27, 2018).

Tanner, J. F. J., Hunt, J. B., & Eppright, D. R. (1991). The protection motivation model:

A normative model of fear appeals. Journal of Marketing, 55(3), 36-45.

Thomas, S. P., Thomas, R. W., Manrodt, K. B., & Rutner, S. M. (2013). An

experimental test of negotiation strategy effects on knowledge sharing intentions

in buyer–supplier relationships. Journal of Supply Chain Management, 49(2),

96-113.

Thompson, J. D. (1967). Organizations in Action. New York, NY: McGraw-Hill.

Thurstone, L. L. (1927). A law of comparative judgment. Psychological Review, 34(4),

273-286.

Tovey, A. (2016). Toyota to halt production after explosion at steel factory. Retrieved

from http://www.telegraph.co.uk/finance/newsbysector/industry/12137595/

Toyota-halts-production-after-explosion-at-steel-factory.html (accessed August

27, 2018).

Tukey, J. W. (1977). Exploratory Data Analysis. Reading, MA: Addison-Wesley.

Tushman, M. L., & Nadler, D. A. (1978). Information processing as an integrating

concept in organizational design. Academy of Management Review, 3(3), 613-

624.

https://www.reuters.com/article/us-japan-quake-toyota/toyota-other-major-japanese-firms-hit-by-quake-damage-supply-disruptions-idUSKCN0XE08O



http://www.telegraph.co.uk/finance/newsbysector/industry/12137595/Toyota-halts-production-after-explosion-at-steel-factory.html

http://www.telegraph.co.uk/finance/newsbysector/industry/12137595/Toyota-halts-production-after-explosion-at-steel-factory.html

References 117

Vanessa, M. S., Jijun, G., & Bansal, P. (2006). Being good while being bad: Social

responsibility and the international diversification of US firms. Journal of

International Business Studies, 37(6), 850-862.

Vanhamme, J., & Grobben, B. (2009). “Too good to be true!”. The effectiveness of

CSR history in countering negative publicity. Journal of Business Ethics, 85(2),

273-283.

Vanhamme, J., Swaen, V., Berens, G., & Janssen, C. (2015). Playing with fire:

Aggravating and buffering effects of ex ante CSR communication campaigns for

companies facing allegations of social irresponsibility. Marketing Letters, 26(4),

565-578.

Vergne, J.-P., & Durand, R. (2011). The path of most persistence: An evolutionary

perspective on path dependence and dynamic capabilities. Organization Studies,

32(3), 365-382.

Vergne, J. P., & Durand, R. (2010). The missing link between the theory and empirics

of path dependence: Conceptual clarification, testability issue, and

methodological implications. Journal of Management Studies, 47(4), 736-759.

Verma, R., Louviere, J. J., & Burke, P. (2006). Using a market-utility-based approach to

designing public services: A case illustration from United States Forest Service.

Journal of Operations Management, 24(4), 407-416.

Verma, R., & Pullman, M. E. (1998). An analysis of the supplier selection process.

Omega, 26(6), 739-750.

Vlachos, P. A., Panagopoulos, N. G., & Rapp, A. A. (2013). Feeling good by doing

good: Employee CSR-induced attributions, job satisfaction, and the role of

charismatic leadership. Journal of Business Ethics, 118(3), 577-588.

Waddock, S. A., & Graves, S. B. (1997). The corporate social performance-financial

performance link. Strategic Management Journal, 18(4), 303-319.

Wagner, T., Lutz, R. J., & Weitz, B. A. (2009). Corporate hypocrisy: Overcoming the

threat of inconsistent corporate social responsibility perceptions. Journal of

Marketing, 73(6), 77-91.

Walker, K., & Wan, F. (2012). The harm of symbolic actions and green-washing:

Corporate actions and communications on environmental performance and their

financial implications. Journal of Business Ethics, 109(2), 227-242.

Walls, J. L., Berrone, P., & Phan, P. H. (2012). Corporate governance and

environmental performance: Is there really a link? Strategic Management

Journal, 33(8), 885-913.

Wangenheim, F. V., & Bayón, T. (2007). The chain from customer satisfaction via

word-of-mouth referrals to new customer acquisition. Journal of the Academy of

Marketing Science, 35(2), 233-249.

Wason, K. D., Polonsky, M. J., & Hyman, M. R. (2002). Designing vignette studies in

marketing. Australasian Marketing Journal, 10(3), 41-58.

Wathne, K. H., Biong, H., & Heide, J. B. (2001). Choice of supplier in embedded

markets: Relationship and marketing program effects. Journal of Marketing,

65(2), 54-66.

References 118

Weaver, G. R., Trevino, L. K., & Cochran, P. L. (1999). Integrated and decoupled

corporate social performance: Management commitments, external pressures,

and corporate ethics practices. Academy of Management Journal, 42(5), 539-

552.

Webb, J. (2016a). Child workers found in clothing supply chain: ASOS, Marks &

Spencer implicated. Retrieved from https://www.forbes.com/sites/jwebb/2016/

10/25/child-workers-found-in-clothing-supply-chain-asos-marks-spencer-

implicated/#6ebce1e04b12 (accessed August 27, 2018).

Webb, J. (2016b). The Taiwanese earthquake that nearly flattened the Apple iPhone 7.

Retrieved from https://www.forbes.com/sites/jwebb/2016/02/29/the-taiwanese-

earthquake-that-nearly-cancelled-the-apple-iphone-7/#36e51f854da7 (accessed

August 27, 2018).

Webb, T. L., & Sheeran, P. (2006). Does changing behavioral intentions engender

behavior change? A meta-analysis of the experimental evidence. Psychological

Bulletin, 132(2), 249.

Weber, E. U., & Milliman, R. A. (1997). Perceived risk attitudes: Relating risk

perception to risky choice. Management Science, 43(2), 123-144.

Weick, K. E. (1979). The Social Psychology of Organizing. New York, NY: McGraw-

Hill.

Westhues, M., & Einwiller, S. (2006). Corporate foundations: Their role for corporate

social responsibility. Corporate Reputation Review, 9(2), 144-153.

Westphal, J. D., & Zajac, E. J. (2001). Decoupling policy from practice: The case of

stock repurchase programs. Administrative Science Quarterly, 46(2), 202-228.

Wickert, C., Scherer, A. G., & Spence, L. J. (2016). Walking and talking corporate

social responsibility: Implications of firm size and organizational cost. Journal

of Management Studies, 53(7), 1169-1196.

Wilcox, K., Kim, H. M., & Sen, S. (2009). Why do consumers buy counterfeit luxury

brands? Journal of Marketing Research, 46(2), 247-259.

Winter, S. G., Cattani, G., & Dorsch, A. (2007). The value of moderate obsession:

Insights from a new model of organizational search. Organization Science,

18(3), 403-419.

Wong, E. (2013). Poultry plant in deadly fire won plaudits from Chinese. Retrieved

from http://www.nytimes.com/2013/06/05/world/asia/chinese-poultry-factory-

was-considered-a-model.html (accessed August 27, 2018).

Wood, D. J. (1991). Corporate social performance revisited. Academy of Management

Review, 16(4), 691-718.

Wotruba, T. R., Chonko, L. B., & Loe, T. W. (2001). The impact of ethics code

familiarity on manager behavior. Journal of Business Ethics, 33(1), 59-69.

Wu, D., & Ting-Fang, C. (2017). Amid hype, iPhone suppliers scramble to ready new

parts. Retrieved from https://asia.nikkei.com/Business/Trends/Amid-hype-

iPhone-suppliers-scramble-to-ready-new-parts (accessed August 27, 2018).

Wu, G. (1999). Anxiety and decision making with delayed resolution of uncertainty.

Theory and Decision, 46(2), 159-199.

https://www.forbes.com/sites/jwebb/2016/10/25/child-workers-found-in-clothing-supply-chain-asos-marks-spencer-implicated/#6ebce1e04b12



https://www.forbes.com/sites/jwebb/2016/02/29/the-taiwanese-earthquake-that-nearly-cancelled-the-apple-iphone-7/#36e51f854da7

https://www.forbes.com/sites/jwebb/2016/02/29/the-taiwanese-earthquake-that-nearly-cancelled-the-apple-iphone-7/#36e51f854da7

http://www.nytimes.com/2013/06/05/world/asia/chinese-poultry-factory-was-considered-a-model.html

http://www.nytimes.com/2013/06/05/world/asia/chinese-poultry-factory-was-considered-a-model.html

https://asia.nikkei.com/Business/Trends/Amid-hype-iPhone-suppliers-scramble-to-ready-new-parts

https://asia.nikkei.com/Business/Trends/Amid-hype-iPhone-suppliers-scramble-to-ready-new-parts

References 119

Yildiz, H., Yoon, J., Talluri, S., & Ho, W. (2016). Reliable supply chain network

design. Decision Sciences, 47(4), 661-698.

Yoon, Y., Gürhan‐Canli, Z., & Schwarz, N. (2006). The effect of corporate social

responsibility (CSR) activities on companies with bad reputations. Journal of

Consumer Psychology, 16(4), 377-390.

Zavyalova, A., Pfarrer, M. D., Reger, R. K., & Shapiro, D. L. (2012). Managing the

message: The effects of firm actions and industry spillovers on media coverage

following wrongdoing. Academy of Management Journal, 55(5), 1079-1101.

Zeelenberg, M., Van den Bos, K., Van Dijk, E., & Pieters, R. (2002). The inaction

effect in the psychology of regret. Journal of Personality and Social Psychology,

82(3), 314-327.

Zott, C., & Huy, Q. N. (2007). How entrepreneurs use symbolic management to acquire

resources. Administrative Science Quarterly, 52(1), 70-105.

Zsidisin, G. A., & Ritchie, B. (2008). Supply Chain Risk: A Handbook of Assessment,

Management, and Performance. New York, NY: Springer.

Curriculum vitae 120

Curriculum vitae

Academic employment

08/2017 – Visiting Researcher

10/2017 Colorado State University, Fort Collins, Colorado, USA

on invitation by Dr. Lynn M. Shore and Dr. John R. Macdonald

08/2015 – Doctoral Candidate/ Research Assistant

09/2018 University of Mannheim, Business School, Mannheim, Germany

Endowed Chair of Procurement

Education

08/2015 – Doctoral Studies in Business Administration (Dr. rer. pol.)


Thesis: ”Decision Making in Supply Risk and Supply Disruption

Management”

Advisor: Prof. Dr. Christoph Bode

09/2013 – M.Sc. in Mannheim Master in Management


09/2012 – ERASMUS Semester abroad

01/2013 NOVA School of Business and Economics, Lisbon, Portugal

10/2010 – B.Sc. in Business Administration

09/2013 University of Cologne, Faculty of Management, Economics, and Social

Sciences, Cologne, Germany

Documents

DECISION MAKING IN SUPPLY RISK AND SUPPLY DISRUPTION